2008-02-01 [16:38:00.0000] http-equiv=page-enter is a hack to make ie not flash when navigating between pages [16:43:00.0000] Good times [17:12:00.0000] http://eric.van-der-vlist.com/blog/2008/01/31/html-5-turns-documents-into-applications/ [17:12:01.0000] /me is confused [17:16:00.0000] "While many people agree that web applications should be designed as documents, HTML 5 appears to propose to move from documents to applications. This seems to me to be a major step… backward!" :) [17:16:01.0000] gotta love the members of the xml crowd who are now desperately trying to back track away from xml requiring fatal handling of errors [17:17:00.0000] (the spec is pretty damn clear about it, it's not easy to argue out of it) [17:18:00.0000] "no, by fatal we meant kill a kitten, not stop processing the document" :P [17:19:00.0000] so i don't really understand what that article is saying, other than apologising for xml, except maybe that html5 should not define interop as much as it does [17:19:01.0000] in which case my response to that feedback would have to be "Thanks but no thanks". [17:19:02.0000] But Hixie, I like having my weblog break every time some jackass posts a comment and forgets to close his

tag. :D [17:22:00.0000] marcosc_: That's not quite sufficient error handling - you'd have to add an OUT_OF_KITTENS_ERR exception to cope with all contingencies [17:22:01.0000] hehe [17:23:00.0000] Ketsuban: You don't want weblog commenters to add boundary could be trouble.. [19:54:00.0000] http://canvex.lazyilluminati.com/misc/parser/tokeniser.html lets you play with a non-document.write-supporting version of the HTML5 tokeniser [19:54:01.0000] wow, my ipod only gets 27/100. not the worst rendering i've seen, though. [19:54:02.0000] oh no [19:54:03.0000] 40/100 [19:54:04.0000] it's just slow around the two perf tests [19:54:05.0000] ipod touch with webkit? [19:55:00.0000] jruderman: yes [19:55:01.0000] yeah, not too bad :) [19:55:02.0000] marcosc_: looks fine to me, except the text goes out of the box [19:55:03.0000] Philip`: may be I should be a good boy and pull up some of those nasty testcases :-) [19:55:04.0000] maybe you should make it indicate when it's doing the perf test [19:55:05.0000] (or pull the perf test from acid3) [19:55:06.0000] hixie, I was wondering if the text going out was ok. So that's ok then. [19:55:07.0000] marcosc_: but that's caused by the wacky font zooming "feature" [19:55:08.0000] marcosc_: well, it's not "ok" per se [19:56:00.0000] marcosc_: but it's intentional on their part [19:56:01.0000] it looks like it's basically a minimum font size [19:56:02.0000] Hixie, thanks for the clarification. [19:57:00.0000] (they couldn't claim to pass like this, but they don't claim to pass acid2, either) [19:58:00.0000] do you consider having a minimum font size by default to be a spec violation? [20:02:00.0000] http://damowmow.com/playground/demos/font-size/a.html suggests it's not actually as simple as it being a minimum font size thing [20:03:00.0000] jruderman: not really, but there are certain cases (like the way gecko lets the min font size pref override even the font size of elements with no text) that i think are wrong [20:03:01.0000] not sure how strongly the spec backs me up [20:13:00.0000] Hixie: heh, I am pretty sure the text-under-iframe-eater is in the tokenizer. uh-oh. [20:14:00.0000] marcosc_: this is really weird behaviour. i don't understand why the text takes so much room on the ipod [20:15:00.0000] Hixie, that's what I thought... which prompted me to ask ;) I thought maybe there was something wrong in the css or something. [20:15:01.0000] that text is definitely bigger than 16px [20:15:02.0000] but i don't understand why [20:17:00.0000] unfortunately, I can't do any testing atm because I don't have wireless here [20:21:00.0000] (grmbl. it explicitly discards stuff within an iframe... Wonder why, it should hide content just fine) [20:28:00.0000] yikes. going to the html5 spec with tokenizer on full debug is not fun [20:31:00.0000] hah [20:31:01.0000] BTW, FF2.0 seems to parse everything inside [23:53:00.0000] ah looks like the current draft actually allows for _top etc: 'The base element can now have a target attribute as well mainly for consistency with the a element and because it was already widely supported. Also, the target attribute for the a and area elements is no longer deprecated, as it is useful in Web applications, for example in conjunction with iframe.' [23:53:01.0000] http://www.w3.org/TR/html5-diff/#new-attributes [02:07:00.0000] /me sees that http://www.angelfire.com/tx/rangerexes/heritage.htm really doesn't work very well in Opera (9.2 or 9.5) [02:08:00.0000] :( [02:08:01.0000] oh, and it's not centered in IE [02:11:00.0000] so Firefox renders everything before
[02:12:00.0000] but it's still centered then [03:51:00.0000] /me thinks more tree construction tests are needed, given that his code is full of 'XXX' and 'TODO' comments but still passes the current test cases [03:53:00.0000] Philip`: Maybe there should be a list of such todos somewhere so it's easy to find out which testcases need writing [03:55:00.0000] I'm just trying to fill in the gaps myself at the moment - if I wrote down a list, I'd forget what anything meant by the time I looked at it again :-) [04:00:00.0000] Philip`, awesome [04:00:01.0000] we wrote our code and later added some tests on top of the 71 hixie made but that's it [04:00:02.0000] then henri added some and we fixed some more bugs, mostly in unicode handling [04:01:00.0000] /me 's implementation only handles ASCII :-( [04:01:01.0000] reminds me of arc [04:12:00.0000] annevk: :) [04:12:01.0000] "SVGSVGTextElement.getNumberOfChars() counts UTF-16 surrogates as separate characters" [04:13:00.0000] mmmm [04:13:01.0000] given that various languages already have surrogates build in, maybe SVG should follow? [04:13:02.0000] The SVGSVG prefix seems accident-prone [04:13:03.0000] (failure for test 69 in Opera) [04:13:04.0000] (in Acid3) [04:43:00.0000] annevk: the DOM is pretty deeply in the territory of counting UTF-16 code units [04:43:01.0000] annevk: and too late to fix [04:48:00.0000] http://james.html5.org/cgi-bin/parsetree/parsetree.py?source=%3C!doctype%20html%3E%3Ctable%3E%3Cp%3E%3C/table%3E - the
should generate implied end tags, which should imply a

, which should cause a table-voodoo error [04:49:00.0000] /me doesn't know if that can have a more significant effect than a lack of error [05:11:00.0000] hsivonen, right, does it make sense for SVG to be different? [05:12:00.0000] annevk: oh it's different? no, that doesn't make sense at all [05:41:00.0000] hmm. is it defined anywhere in the html5 spec that elements implement appropriate interfaces when put in the dom by the xml processor? or elements created with createElementNS? [05:42:00.0000] i note that the html5 parser says that elements should implement the appropriate interfaces [05:43:00.0000] perhaps it should be defined in a higher layer? like "when an element is created and it is in the html namespace, it must implement the appropriate interfaces..." or some such [05:45:00.0000] zcorpan_, since HTML5 is defined in terms of the DOM, it doesn't matter where the elements come from, they still implement the DOM APIs [05:46:00.0000] Lachy: so why is there a requirement in the parser about elements implementing appropriate interfaces? [05:46:01.0000] I don't know [05:46:02.0000] seems kind of redundant [05:46:03.0000] zcorpan_: I think the spec needs to say that XML stuff needs to implement the right stuff [06:48:00.0000] hsivonen, is there a summary somewhere of how DOM and ECMAScript assume UTF-16? [07:13:00.0000] defining it at an intermediate layer seems cleaner to me [08:18:00.0000] /me lets his frustration out at the httpbis wg [08:21:00.0000] [08:23:00.0000] /me now has 2000 lines of generated JavaScript to implement a tree constructor [08:24:00.0000] though unfortunately most of the lines say things like debug("TODO: ReprocessAsIf"); because I haven't implemented them yet :-( [08:25:00.0000] gsnedders: User-Agents MUST ignore the user preferences? That's a nice one. [08:25:01.0000] /me wonders whether someone has stats on wrong content-type w/in http headers. He surely has seen it with http-equiv [08:25:02.0000] SadEagle: most of what the HTTPbis WG is working on is totally idealistic and backwards incompatible (well, compatible with conforming impls. of RFC2616, if there are any) [08:26:00.0000] SadEagle: content-type or charset? [08:26:01.0000] SadEagle: charset's not that widespread. The issue is with implicit charsets (like on text/xml) [08:27:00.0000] I meant charset within content-type. [08:27:01.0000] yeah, if it is explicitly given, it's normally right (except for one or two things, like ISO-8859-1 and GB2302) [08:28:00.0000] (not GB2302, GB2312) [08:29:00.0000] text/xml doesn't have explicit charsets [08:29:01.0000] see RFC 3023 [08:29:02.0000] (within the file, that is) [08:29:03.0000] /me -> food [08:30:00.0000] annevk: I mean on the actual MIME type [08:30:01.0000] (i.e., parameters on the MIME type [08:31:00.0000] well, I've seen e.g. .... encoded as ASCII. Also koi8/cp1251 messups. Now, w/meta one can claim it's html that defines it anyway, but still, I would expect the sequence in case of a misconfigured server to be: [08:31:01.0000] SadEagle: that's specific to HTML, though. I'm talking in general HTTP terms. UTF-16 in meta@charset must be ignored [08:32:00.0000] 1) User open a webpage, sees garbage. 2) User activates encoding auto-detection, which picks the right decoding. 3) The user keeps browsing the webpage with auto-detection (or manualyl chosen charset) "sticky", and has no problem. It seems to me that the proposal would require the user to repeat (2) every time they navigate. [08:33:00.0000] yeah, it would [09:04:00.0000] hsivonen: v.nu says that

is an html4-only error [09:34:00.0000] hsivonen: it seems that your encoding declaration warnings have gone amok: http://validator.nu/?doc=http%3A%2F%2Fvalidator.w3.org%2F [09:38:00.0000] webben: target=_top is allowed, it's only _blank that isn't. [09:39:00.0000] /me notes that v.w.o has "Group Error Messages by type" [09:39:01.0000] Hixie: Thanks, I did note that. :) [10:20:00.0000] zcorpan_: oops. thanks [10:22:00.0000] /me is feeling very naïve at the moment [10:22:01.0000] /me knows nowhere near enough about Python [10:38:00.0000] zcorpan_: fixed. thanks [11:52:00.0000] jwalden: yt? [11:53:00.0000] Hixie: sorta; you caught me just before a phone conference actually :-) [11:54:00.0000] i just made a change to MessageEvent/postMessage() for security that fixed a separate problem from the one that was raised earlier [11:54:01.0000] i'm about to send mail to whatwg explaining it [11:54:02.0000] annevk: see above also [11:54:03.0000] basically changing message.domain and message.uri to just message.origin [11:54:04.0000] which is what the .uri value used to be, but without the path [11:55:00.0000] hm [11:55:01.0000] I await the explanation! [11:55:02.0000] :-) [11:57:00.0000] sent [12:12:00.0000] nice catch on username/password; I should have thought of that in my testing [12:12:01.0000] I *did* think of it once you summarized the change above, tho :-) [12:12:02.0000] :-) [12:12:03.0000] i think this, as a sideeffect, will simplify addressing one of the other problems [12:13:00.0000] instead of postMessage(message, [domain, [uri]]); we can just have postMessage(message, [origin]); [12:13:01.0000] so everything before the path, *except* for username/password, then [12:13:02.0000] yeah [12:13:03.0000] (and later make it postMessage(message, [endPoint], [origin]), since those two are type-distinguishable) [12:29:00.0000] Hixie: what commit did this change occur in? I can't find it in the tracker [12:29:01.0000] I sent my response [12:30:00.0000] /me is writing a bug report for WebKit and wants to link to the change [12:30:01.0000] I'll note that I originally wanted .domain to always include a port number, which would have avoided the first problem even if I'd never have noticed that we'd avoided it :-P [12:30:02.0000] and I definitely wouldn't have noticed the first problem [12:31:00.0000] I'm still a bit chagrined I missed the second [12:31:01.0000] GAAAAH [12:32:00.0000] /me gets bitten by forgetting +whatwg AGAIN [12:36:00.0000] /me bits jwalden [12:36:01.0000] ? [12:36:02.0000] *bites [12:37:00.0000] jgraham: I took a good look at the treebuilders today, and I really can't work out how the etree stuff works. [12:37:01.0000] gsnedders: By magic :) [12:38:00.0000] :) [12:38:01.0000] (i.e., I can't get DOM to work in a similar way) [12:38:02.0000] What do you actually want to know? [12:38:03.0000] I just don't really understand how the etree stuff works _at all_. [12:39:00.0000] as in, how it works out what etree impl to use from the second param to getTreeBuilder [12:40:00.0000] So, from memory, we do something like create a module-like object at runtime with a module-level variable ElementTree set to the module of the Element Tree implementation that we're using [12:41:00.0000] what does getETreeBuilder() do? [12:41:01.0000] /me decides looking t the code would help [12:41:02.0000] what's **kwargs? [12:42:00.0000] heck, what does *arg and **arg mean? [12:42:01.0000] it's different ways of passing arguments in Python [12:42:02.0000] IIRC kwargs is named parameters [12:42:03.0000] can't remember the difference between *arg and **arg [12:43:00.0000] *args means take the tuple args and expand its value as a set of arguments to a function [12:43:01.0000] ah [12:44:00.0000] so something like func(*(True, 3)) calls func with the args True and 3 [12:44:01.0000] and ** is reference? [12:44:02.0000] similarly **kwargs expands a dict kwargs as a set of named parameters to a function [12:44:03.0000] ah [12:44:04.0000] Hixie: you didn't update initMessageEvent(NS)? [12:44:05.0000] then how do you pass by reference? [12:45:00.0000] so func(**{"foo":True, "bar":3}) == func(foo=True, bar=3) [12:45:01.0000] gsnedders: You don't [12:45:02.0000] yeah. that makes sense. [12:45:03.0000] jgraham_: ah. [12:45:04.0000] jgraham_: that's simple. [12:45:05.0000] or rather, you always pass by _object_ [12:45:06.0000] now, back to treebuilders [12:48:00.0000] so something like "a = [1,2]; func = lambda x:x.append(3);func(a);print a" will print [1,2,3] [12:48:01.0000] ya [12:51:00.0000] gsnedders: So, getETreeBuilder is like a function to return a set of functions which are all closures over the value of ElementTreeImplementation [12:52:00.0000] notice the return locals() at the end [12:52:01.0000] those functions are turned into a module on the fly in getEtreeModule [12:53:00.0000] This is kinda icky but I wasn't sure how else to reuse all the code for multiple etree implementations [12:53:01.0000] /me has to go eat now, will be back later [12:54:00.0000] Hixie: you have in html5, what's this meant to be? [12:54:01.0000] (or is my in-head parser not got some quirky error handling?) [12:55:00.0000] Hixie: (see "

For the purposes of the interaction of HTML with Selectors' /me throws this proposal for recreating toc(1) and num(1) out the window for being far too expensive [13:04:00.0000] Hixie: that is per HTML 5 seemingly an empty title"" attribute. I guess you didn't mean that :) [13:15:00.0000] Is there some way I can construct a DOM element using JS so it has an attribute called something like title"" without throwing exceptions? [13:16:00.0000] I can't make a very good HTML parser if there's no way to construct the things that are parsed :-( [13:16:01.0000] can you use String.fromCharCode? [13:17:00.0000] The problem is just when trying to pass the string into setAttribute [13:17:01.0000] it throws an exception in that case? [13:18:00.0000] In Firefox 2, yes [13:18:01.0000] (In Opera 9.5, no) [13:18:02.0000] (In Safari 3, yes) [13:19:00.0000] presumably because it's a totally bogus attribute name [13:23:00.0000] It should. [13:24:00.0000] gsnedders: Did any of that help? [13:24:01.0000] jgraham_: I'm still in the middle of finishing off an email :P [13:24:02.0000] (ironically, this one email is probably longer than the total amount of writing I've done at school over the past two days) [13:30:00.0000] hm [13:31:00.0000] isn't this domain hazard also common to document.domain as well? [13:31:01.0000] OK, now what jgraham_ said [13:51:00.0000] jwalden: as i understand it, doc.domain doesn't allow you to cross ports or schemes [13:52:00.0000] Hixie: I could have sworn Mozilla did differently last time I looked [13:52:01.0000] ah [13:56:00.0000] jgraham_: OK, that makes sense now. [13:56:01.0000] (no, it didn't take 20 minutes, I went and procrastinated) [13:57:00.0000] From the looks of things, it'd be simplest to just do the same as etree does [13:57:01.0000] Though it looks like the only thing that'll change for each DOM impl is the TreeBuilder class itself [13:59:00.0000] Or rather, that's the only one that needs to reference the actual impl. [14:11:00.0000] Has any browser implemented datagrid already? [14:11:01.0000] /me is trying nigthtly build without success [14:13:00.0000] met_: not as fr as I know [14:13:01.0000] s/fr/far/ [14:14:00.0000] /me wonders how much effort it would be to implement in js [14:14:01.0000] thx, it looks same for me [14:14:02.0000] preparing some examples for html5 presentation and some datragrid example should be nice and powerfull 8-) [14:15:00.0000] but I have nowhere to check it [14:18:00.0000] jgraham_: would you say just chucking everything in dom.py into a function, like etree.py, would be the best way to go about it? [14:19:00.0000] brb [14:20:00.0000] gsnedders: I don't like to make claims about best :) But consistency is good and if we decide the whole design sucks later it's easier to change two similar things than two different things [14:20:01.0000] Philip`: no, you can't (re title"") [15:00:00.0000] jgraham: I'll try and impl it tomorrow [15:41:00.0000] Hixie, as far as I can tell event listeners are asynchronous which means that the UA could decide to not run process them directly but instead process some code further on... [15:45:00.0000] event listeners are synchronous. [15:45:01.0000] as in, when you dispatch an event, everything from that point to when it returns is synchronous [15:45:02.0000] see dom2 events [15:45:03.0000] i guess i'll admit defeat [15:45:04.0000] /me wonders why everyone fails [15:50:00.0000] if nobody allows click() to nest, i can remove the test [15:50:01.0000] iirc, at least one browser nested at least once [15:51:00.0000] could be [15:54:00.0000] /me sees that http://blogs.msdn.com/ie/archive/2008/01/21/compatibility-and-ie8.aspx has reached exactly 666 comments [15:55:00.0000] It seems a bit odd that it's generated more noise than the Acid 2 post [15:55:01.0000] why? [15:55:02.0000] microsoft has an image problem. people aren't going to be pro-ms until they have proven over years of not being evil that they are in fact good. [15:56:00.0000] acid2 is a small step in the right direction. the compatibility meta header is a giant step backwards. [15:56:01.0000] I love the suggestion in the thread that any unknown doctype be rendered as XHTML 1.1. I think I'll stop reading the noise now. [15:58:00.0000] Hmm, okay, maybe it's not actually odd :-) 2008-02-13 [16:01:00.0000] Hixie, should access control and postMessage() align formats for origin etc? [16:01:01.0000] (actually, the answer is yes, so we should probably resolve it...) [16:03:00.0000] scheme://hostname:port [16:03:01.0000] it's just the uri format without a path [16:05:00.0000] the differences were that your proposed format included a trailing slash and omitted the default port [16:05:01.0000] from a very brief scan [16:05:02.0000] /me hasn't actually checked the spec though [16:05:03.0000] /me just the e-mail [16:07:00.0000] sees that the spec doesn't have the trailing slash but does omit the default port [16:08:00.0000] yeah, i don't figure most people will include the port [16:09:00.0000] ok, I suppose I could make that change to Access-Control-Origin [16:10:00.0000] access-item is already flexible enough (does not allow / though) [16:10:01.0000] the / is part of the path [16:10:02.0000] right [16:10:03.0000] i was wondering whether to allow a trailing slash for copy & paste friendlyness [16:11:00.0000] well, postMessage() actually allows the entire URI [16:11:01.0000] it just ignores the path [16:11:02.0000] and it doesn't include it in event.origin, though that's poorly defined right now [16:12:00.0000] k, i'm not going to change that for access-item [16:15:00.0000] Hixie, maybe you should use the same syntax as access-item... [16:16:00.0000] annevk: elaborate? [16:18:00.0000] for the second argument of postMessage() [16:18:01.0000] i mean, elaborate on what you would want changed in the syntax :-) [16:18:02.0000] right now, it's just a plain old uri/iri [16:18:03.0000] so you can use postMessage("...", "example.org") [16:19:00.0000] oh [16:19:01.0000] why? [16:19:02.0000] that just seems like asking for confusion... [16:19:03.0000] i guess i'd recommend that access-item be simplified too :-) [16:21:00.0000] hmm [16:22:00.0000] (in particular, i may just make postMessage("...", "example.org") resolve the uri locally instead of throwing an exception) [16:23:00.0000] oh, and postMessage("...", "*.example.org") would work [16:24:00.0000] that would make it way more complex [16:24:01.0000] probably not so relevant for postMessage, indeed [16:24:02.0000] 99% of the time, you'll just do postMessage("...", e.origin); [16:24:03.0000] to prevent sending messages to new hosts [16:24:04.0000] and on the long run, you'll use endPoints, anyway [16:25:00.0000] and just use postMessage() to set up the original connection [16:28:00.0000] suggestions for simplifying access-item are welcome on public-appformats btw :) [16:29:00.0000] Hmph, Opera (9.5) apparently doesn't let me add a doctype to a document [16:29:01.0000] Opera only supports ninja doctypes [16:30:00.0000] Uhhh [16:33:00.0000] /me makes it add a comment instead of a doctype [16:37:00.0000] /me -> bed [16:37:01.0000] Also, Opera deletes "--" from inside comments [16:39:00.0000] http://software.hixie.ch/utilities/js/live-dom-viewer/?%3C!DOCTYPE%20html%3E%0D%0A%3Cbody%3E%0D%0A%3Cscript%3Edocument.body.appendChild(document.createComment('a-b--c---d----e'))%3C%2Fscript%3E [16:39:01.0000] Looks like an Opera bug [16:52:00.0000] does anyone have anything they'd like me to prioritise? [16:56:00.0000] No, but thank you for asking :-) [16:58:00.0000] Someone should make something like JSON that can handle 'var x=[];x[0]=x' [17:22:00.0000] Philip`: yaml can probably do that [17:25:00.0000] I think YAML has "&label" and "*label" to mark and refer to bits of structure when you want non-trees, and it seems to be quite handy when writing by hand [17:25:01.0000] (I've no idea how useful or ugly it is when automatically serialising, though) [17:29:00.0000] (But YAML is too crazy and complex to be used where interoperability matters - the spec is a hundred pages long) [17:31:00.0000] so should U+0085 NEXT LINE (NEL) be a space character in html5? [17:31:01.0000] i'm thinking no. [17:33:00.0000] I'd like no because then I could pass UTF-8 through my ASCII parser and it'd still work alright [17:34:00.0000] /me is randomly replying to feedback from random folders [17:36:00.0000] Pick me, pick me! [17:37:00.0000] which folder are you? :-) [17:37:01.0000] does dir="" always default to ltr? [17:37:02.0000] Dunno, I just felt like saying it [17:37:03.0000] or is there a UA pref for the default? [17:46:00.0000] Hixie: I would suspect that that's based on locales [23:50:00.0000] Hixie: re: NEL, please no [23:51:00.0000] Reason: non-ASCII syntax-significant chars => bad [00:16:00.0000] Hixie, parsing could use an update [00:29:00.0000] webben++ http://www.w3.org/mid/47B1B887.7040205⊙gc [03:07:00.0000] the DOM is so bloated :( [03:07:01.0000] it's pretty sad [03:07:02.0000] mostly because it's hard to change at this point [03:10:00.0000] the DOM is very sad. [03:11:00.0000] but the Attribute node thing might be changeable at this point [03:11:01.0000] also getAttribute when the attribute is absent [03:11:02.0000] so I was thinking about that [03:11:03.0000] it seems to me that Attr is important because it has namespaceURI [03:12:00.0000] so can we really change that? [03:12:01.0000] do you mean enumerating the attributes instead querying one? [03:13:00.0000] yeah [03:16:00.0000] though maybe Attr could be changed in a way that it does not inherit from Node [03:17:00.0000] because Note is fricking bloated [03:17:01.0000] s/Note/Node/ [03:18:00.0000] it would be interesting to know if Attr is in practice used by any scripts except serializers [03:18:01.0000] do Ajax libraries do that kind of looping over all attributes? [03:22:00.0000] element.attributes is used I think [03:56:00.0000] hsivonen, btw, do you have a list of things where the DOM assumes UTF-16? [03:57:00.0000] annevk: I don't. [03:58:00.0000] annevk: perhaps saying that the DOM requires it misses the point. it's JavaScript strings that assume 16-bit units [03:59:00.0000] DOM Range is a place where utf-16 is assumed [03:59:01.0000] thanks [04:00:00.0000] i think i'll raise this with the SVG WG [04:00:01.0000] Opera does "UTF-16 units" for getNumberOfChars() and I think it makes sense to keep it that way [04:00:02.0000] and not have a slightly different API just for SVG [04:15:00.0000] hmm, Zs doesn't include u0009–u000D [04:22:00.0000] http://lists.w3.org/Archives/Public/www-svg/2008Feb/0017.html [05:21:00.0000] was there someone on this channel working on Comet stuff? [05:22:00.0000] at some point [05:39:00.0000] man, who uses nodeiterator anyway? [05:39:01.0000] complex :( [05:41:00.0000] and not just a little bit [06:03:00.0000] should OPTIONS on a non-existent URI path return 404? [06:04:00.0000] presumably yes [06:06:00.0000] as in, a non-existent GET URI path? [06:08:00.0000] annevk: yes [06:08:01.0000] annevk: if the URI would be 404 on GET or POST [06:09:00.0000] either that or maybe METHOD NOT ALLOWED (405 or so?) [06:11:00.0000] the method is allowed thoguh [06:11:01.0000] *though [06:11:02.0000] so that status code can't be used [06:30:00.0000] Hixie, where in DOM Level 2 Traversal is it defined what to do with incorrect return values for filters? Such as 'true' and 'false'? [06:36:00.0000] something tells me Dmitry Turin hasn't paid good attention to what others (Hixie in particular) have written on public-html [06:37:00.0000] i always forget to bookmark those posts for reuse as I expect people to understand... [07:45:00.0000] I wonder if access-control is *now* stable enough to implement in Validator.nu... [07:49:00.0000] annevk: surely the GET algorithm should apply to HEAD as well [07:49:01.0000] in access-control [07:53:00.0000] I'm not sure how I far I want to open that door [07:54:00.0000] hmm. looks like Gecko source now has OPTIONS but still the Method-Check stuff [07:54:01.0000] /me implements per spec [07:58:00.0000] hsivonen, I'd wait for Firefox 3 to ship [07:59:00.0000] annevk: do you expect changes to the single-URI policy version? [08:04:00.0000] "from xxx import *" works fine as a normal import in Python, but how can you deal with such a thing when the module is bound to a variable? [08:05:00.0000] /me wonders why #dom li:last-child::before doesn't work right in Firefox [08:05:01.0000] jgraham__: maintaining backwards compat. is harder than I thought, because everything refers to specific classes/functions in the dom treebuilder [08:09:00.0000] Oh, it does work right in Firefox once you make it reflow or something [08:17:00.0000] hrm. OPTIONS * seems like a useless complication that doesn't have an upside [08:20:00.0000] annevk: access-control could use a companion guide for server-side devs [08:33:00.0000] hsivonen, I hope on no changes whatsoever [08:33:01.0000] however, I have hoped that for the past year [08:37:00.0000] hsivonen, OPTIONS *? I'll remind you that you argued for using OPTIONS too :) [08:38:00.0000] annevk: yeah, OPTIONS is the right way to go, but OPTIONS * is still an unnecessary complication in RFC 2616 [08:39:00.0000] annevk: I've now deployed access-control on v.nu [08:39:01.0000] have fun [08:40:00.0000] annevk: but it seems that OPTIONS * stuff never reaches the servlet [08:40:01.0000] not a problem. just weird [08:40:02.0000] and useless [08:41:00.0000] oops. And I broke html5.validator.nu and parsetree.validator.nu, it seems [08:42:00.0000] that promises something for when less competent people start doing this :( [08:43:00.0000] annevk: that's not related to handling options. it is another "fix" in the same checkout [08:57:00.0000] fixed [08:58:00.0000] /me sighs, thinking his change to html5lib is finally working [08:58:01.0000] Now, to commit. [09:00:00.0000] /me isn't sure if he ever has actually commited before, even though he's had perms for a while [09:22:00.0000] so the main reasons people use Transitional seems to be (1) be able to use target=_blank and (2) using software that emits markup that only validates as transitional (also mainly due to target=_blank) [09:54:00.0000] zcorpan: Isn't iframe also a factor? [09:55:00.0000] (a major factor, I mean) [09:56:00.0000] webben: less so than target=_blank, it seems. however, iframe is already valid html5 [09:56:01.0000] zcorpan: is this based on 10 answers to the sitepoint survey? [09:57:00.0000] hsivonen: yes, and the comments [09:59:00.0000] s/people/people who are on sitepoint and read the thread and care enough to vote or comment/ [10:08:00.0000] ergh. [10:08:01.0000] html5lib relies on minidom's brokenness. [10:10:00.0000] http://philip.html5.org/tools/parser/ [10:11:00.0000] Currently missing quite a few bits (especially anything to do with CDATA or RCDATA etc) but generally it seems to work [10:11:01.0000] Philip`: that your JS impl? [10:11:02.0000] Yes [10:14:00.0000] — why do we set a readonly property? [10:16:00.0000] It's irritating that the DOM interface doesn't let you create as many DOMs as an XML parser or HTML parser can [10:17:00.0000] Philip`: huh? [10:17:01.0000] we set that readonly property to work around a bug in minidom, seemingly [10:20:00.0000] Oh, I suppose for ownerDocument you can just pass the doctype into createDocument and it'll work alright, so maybe you can do as much as an XML parser [10:20:01.0000] (but it's still not possible to write a proper HTML5 parser in JS because the DOM won't let you create elements and attributes with funny names) [10:22:00.0000] Philip`: that's really useful [10:23:00.0000] /me wonders how else to work around that bug in minidom [10:23:01.0000] gsnedders: Why does the current workaround not work? [10:24:00.0000] Philip`: are you catching invalid names for attributes? [10:24:01.0000] Philip`: if you aren't using minidom, and you're using something that enforces the readonly-ness of the property, it breaks [10:24:02.0000] How can I check if "Dom" is the module xml.dom.minidom? [10:24:03.0000] gsnedders: If the minidom bug doesn't affect any other DOMs, you could perhaps wrap the property-setting line in try/except and ignore failures [10:25:00.0000] Philip`: the exception raised depends on what impl you're using, though [10:27:00.0000] Philip`: is handled correctly? [10:27:01.0000] zcorpan: Oops, I wasn't checking attributes - fixed now [10:28:00.0000] zcorpan: Depends on what you mean by "correctly" [10:28:01.0000] Philip`: local name "foo:bar" rather than "bar" [10:29:00.0000] zcorpan: some DOM's don't allow attributes to be created containing with a ":" [10:29:01.0000] zcorpan: It just does createElementNS('http://www.w3.org/1999/xhtml', 'foo:bar'), which seems to work with localName correctly [10:29:02.0000] gsnedders: the w3c dom does [10:30:00.0000] s/'// [10:30:01.0000] Philip`: i think that ends up being a "bar" element [10:30:02.0000] s/work/not work/ [10:30:03.0000] I agree now :-) [10:30:04.0000]

gets parsed wrongly [10:30:05.0000] (because my code just tests localName) [10:30:06.0000] zcorpan: I get a "foo:bar" in Saf3 [10:31:00.0000] /me wonders how he should fix that [10:32:00.0000] Philip`: either createElement() or throw when there are colons [10:32:01.0000] zcorpan: Could createElement work when I want these things to be in the HTML namespace? [10:33:00.0000] Philip`: if the browser puts them in the html namespace then yes :) [10:35:00.0000] Why would it put them in the HTML namespace, when they're not being added into an existing HTML document? [10:36:00.0000] ah. hmm. perhaps because dom5 core will require it? :) [10:36:01.0000] document.implementation.createDocument('a', 'b', null).createElement('foo:bar').namespaceURI only works in Opera :-( [10:40:00.0000] Philip`: how come

and are different? [10:48:00.0000] zcorpan: doesn't get added to the list of active elements, since it's not an "a" start tag, and so the doesn't realise there's already an active "a" [10:48:01.0000] where does get added to the stack of open elements, and so the

does think there's already a "p" [10:51:00.0000] zcorpan: I've updated it to distrust anything that's not /^[a-zA-Z_][a-zA-Z0-9_.-]*$/, which will hopefully avoid those problems [10:53:00.0000] Philip`: that seems a bit harsh [10:53:01.0000] are there other characters than the colon that are problematic? [10:54:00.0000] zcorpan: Anything that's not an XML Name should cause an exception [10:54:01.0000] Philip`: yeah, but it did that before too, no? [10:55:00.0000] (modulo browser bugs) [10:55:01.0000] It used to cause an exception and then catch it and get rid of anything that's /[^a-zA-Z_]/ [10:55:02.0000] (except in browsers that failed to throw the exception) [10:56:00.0000] (in which case it'd act inconsistently with other browsers, which is probably undesirable) [10:57:00.0000] well, i guess there are few valid reasons to use anyway [10:57:01.0000] Or maybe I should just use createElement and forget about the namespace thing, at least for the DOM-viewer version [10:57:02.0000] sounds reasonable [10:58:00.0000] except that shouldn't help because createElement and setAttribute should still throw exceptions if it's not an XML Name [10:58:01.0000] it solves the colon problem [10:58:02.0000] (and the only difference is it doesn't have to be a QName) [10:59:00.0000] and that createElement() takes local name as input rather than qname [11:02:00.0000] zcorpan: Okay, I've changed it to just use createElement by default, and to be more permissive about characters [11:04:00.0000] (It accepts /^[a-zA-Z:_][a-zA-Z0-9:_.-]*$/, else it gets rid of all /[^a-zA-Z_]/) [11:04:01.0000] why not check against the Name production? [11:04:02.0000] Because that seems long and complex and not very useful [11:05:00.0000] ok, why not let the browser throw then? [11:06:00.0000] Because then it'll give different output in different browsers (depending on the browser's bugs), which will be a little confusing [11:09:00.0000] but testing nonascii or random junk is useful [11:10:00.0000] /me is off for food [11:10:01.0000] "Either they continue to use HTML4 Transitional, or they stop to care about conformance, or they use [11:10:02.0000] scripts to insert the target attribute or use window.open(). [11:10:03.0000] " [11:10:04.0000] the last two options are the same as the second option [11:11:00.0000] "the stop to care about the output of conformance checkers" [11:11:01.0000] s//y/ [11:12:00.0000] conformance checkers should check script too [11:13:00.0000] would be interesting to write a conformance checker that supported rewinding [11:13:01.0000] that just went down every branch [11:13:02.0000] checking arguments to every method, etc [11:14:00.0000] Bonus points if it doesn't take infinite time [11:14:01.0000] well obviously it wouldn't check everything [11:15:00.0000] but you could catch a number of simple things [11:16:00.0000] So some well-intentioned person will write some obfuscated code that opens windows in a way the conformance checker can't detect, and then other well-intentioned people will copy-and-paste that so that their pages validate successfully [11:16:01.0000] and they'll get a warning "warning: script could not be verified" [11:17:00.0000] if people want to ignore a validator, why would they use a validator? [11:17:01.0000] They'll ignore those warnings because every non-trivial script will give that warning [11:17:02.0000] i think most people don't realise that setting target from script is non-conforming [11:19:00.0000] It's hard enough trying to decide that some code will always terminate, without always giving up and saying "can't verify code" [11:24:00.0000] i'm just sayin', it'd be useful [11:24:01.0000] Many impossible things would be useful :-) [11:24:02.0000] s/impossible/hard/ [11:25:00.0000] annevk: just to update you on what olliej mentioned last night [11:25:01.0000] annevk: I am in the process of implementing the new postMessage [11:26:00.0000] annevk: it has not landed yet though [11:26:01.0000] Hixie: can you fix the occurrence of title"" in source? [11:38:00.0000] gsnedders: i thought i had [11:39:00.0000] Hixie: oh, wait. I doing this from a local copy. [11:39:01.0000] /me looks at the current copy [11:39:02.0000] yeah, it's fixed. [11:39:03.0000] k [11:40:00.0000] /me is too tired [11:41:00.0000] Hixie: for the cross-referencing (per HTML 5) should I do any normalisation (of whitespace? of Unicode? or case?) before calculating cross references? [11:42:00.0000] space characters should go to one space, yeah [11:42:01.0000] several of the title attributes and element contents span multiple lines [11:42:02.0000] currently the spec says it must be exactly identical [11:43:00.0000] Hixie: (I mean in the HTML 5 def of dfn) [11:43:01.0000] probably best to normaline case too [11:43:02.0000] yeah that's all gonna be dropped i think [11:43:03.0000] ah. [11:46:00.0000] /me notes that on his private playing around copy that actually processing the DOM is now slower than parsing [11:47:00.0000] import psyco [11:47:01.0000] Makes things much faster! [11:48:00.0000] /me ponders [11:48:01.0000] /me expands graph [11:48:02.0000] Okay, that isn't good. [11:49:00.0000] Python is definitely too slow. That sucks. [11:49:01.0000] You could use Java [11:50:00.0000] I know no Java, which is a slight issue. [11:50:01.0000] python is slow, generally, but i wouldn't bet on python's speed being the gating factor unless you're making a ray tracer or something [11:51:00.0000] Hixie: or maybe `cat header_whatwg source` is too big for a python spec-gen :) [11:51:01.0000] shouldn't be [11:51:02.0000] i mean, the overhead isn't going to be _that_ big, regardless of what language you use [11:52:00.0000] (the spec splitter is a python script, e.g.) [11:53:00.0000] Hixie: ok, s/conformance/validating their documents/ [11:54:00.0000] /me thinks he could try doing this another way, actually [11:54:01.0000] Hixie: also, window.open() isn't non-conforming, is it? [11:54:02.0000] zcorpan_: window.open with _blank is [11:54:03.0000] iirc [11:54:04.0000] window.open(url) works [11:55:00.0000] iirc that's non-conforming [11:55:01.0000] oh, i guess it's not explicitly non-conforming [11:55:02.0000] oh well [11:55:03.0000] anyway i think you're right, and _blank will just have to be made valid. we'll just have to make it optional to implement instead. [11:56:00.0000] /me thinks he'll be able to do it in a more memory expensive way [11:56:01.0000] makes sense [11:56:02.0000] (but the overhead is still not that big) [11:56:03.0000] zcorpan_: any time the spec requires opening a new top-level browsing context, i think the current one should be selected instead [11:56:04.0000] zcorpan_: i'm getting more and more annoyed by sites that decide when i'm opening a page in a new tab [11:57:00.0000] and since all i have is a spec-hammer, all problems look like spec-nails to me [11:57:01.0000] Hixie: yeah, but that's configurable in the browser [11:57:02.0000] Hixie: weinig has a question for you (and he's a pansy) [11:57:03.0000] not currently, but yeah [11:57:04.0000] bradee-oh: brt [11:58:00.0000] Hixie: using the spec-hammer doesn't stop sites opening new windows, it just makes it harder to configure because it becomes an arms race [11:58:01.0000] jgraham__: ping [11:58:02.0000] zcorpan_: yeah [11:58:03.0000] Hixie: hey, we were just curious what an refers to? [11:59:00.0000] weinig: same as , but for IRIs [11:59:01.0000] Hixie: ah, ok [11:59:02.0000] weinig: did i use it in a paragraph without a reference? [11:59:03.0000] Hixie: in Resource metadata management [12:00:00.0000] oops [12:00:01.0000] someone send mail :-) [12:00:02.0000] weinig: can you? [12:00:03.0000] gsnedders: sure [12:01:00.0000] Hixie: I will also send mail regarding this, but, is postMessage supposed to work with file: urls? [12:01:01.0000] (i.e., not the guy who pointed out where it was, but the one who found it) [12:01:02.0000] weinig: basically, anything involving file: URIs is not my concern, since you don't need interop for anything that's using file: URIs. so it's up to you. [12:01:03.0000] weinig: but generally i would say i guess it would work, but origin would be null [12:02:00.0000] /me nods [12:02:01.0000] Hixie: when replacing spaces for x-refs, what space characters should be done? just ASCII ones? [12:04:00.0000] i'd just do whatever falls under the "space character" definition in html5 [12:04:01.0000] k. [12:04:02.0000] but for me, it just needs to be space and linefeed [12:04:03.0000] all "me" Hixie :) [12:13:00.0000] /me doesn't particually want to commit this to html5lib because he expects he'll be shot for it [13:03:00.0000] gsnedders: I'm kinda here [13:04:00.0000] jgraham__: would I be shot if I committed a "except Exception"? [13:04:01.0000] specifically, to work around a bug in minidom [13:04:02.0000] No, well you would unless you just write except: (no need for the Exception bit) :) [13:05:00.0000] gsnedders: Yeah, that seems like a reasonable use case [13:05:01.0000] (I read the logs earlier) [13:05:02.0000] the exact exception depends on the DOM impl, as I said [13:06:00.0000] Yeah, the alternative (if Dom == xml.dom.minidom) seems just as bad [13:10:00.0000] jgraham__: and should I add pxdom to the parser tests even though one test fails, due to a bug in pxdom? [13:11:00.0000] gsnedders: Yeah, for sure. But email html5lib-discuss⊙gc explaining why the test fails [13:11:01.0000] k [13:11:02.0000] (I've already emailed the author of pxdom about the bug, fwiw) [13:15:00.0000] gsnedders: Have you looked at making multiple dom imps work with treewalkers [13:16:00.0000] jgraham__: I haven't yet tried, but a quick look at the code made me think it would work [13:16:01.0000] Yeah, I don't know how much woulkd need to be changed. Maybe just adding pyxdom to the tests [13:17:00.0000] jgraham__: what file would they need to be added in for that? [13:17:01.0000] test_treewalkers.py? [13:18:00.0000] tests/testtreewalkers.py [13:18:01.0000] yeah. [13:19:00.0000] thanks weinig [13:22:00.0000] jgraham__: yeah, it works fine with the DOM treewalker. [13:22:01.0000] (pending the other bug that comes up again) [13:22:02.0000] Great [13:29:00.0000] I think I'll send a general email about what I've done [13:41:00.0000] /me commits and sends email [13:45:00.0000] updated http://www.hixie.ch/specs/dom/messages/0.9 [13:48:00.0000] could you add some rational etc? [13:49:00.0000] it's not really clear to me what problem it's trying to solve [13:51:00.0000] rationale, you mean? [13:51:01.0000] sure [13:52:00.0000] yeah [14:28:00.0000] it would be interesting to have a .reply() API for the postMessage() stuff [14:39:00.0000] Hixie: in, e.g., step 2.4 of the postMessage() algorithm, does "these steps" mean the substeps of step 2 or all the steps of the algorithm? [14:39:01.0000] Hixie: I assume the latter [14:45:00.0000] all of them 2008-02-14 [16:29:00.0000] well bummer [16:29:01.0000] /me ends up speccing an event that fires on garbage collection [16:29:02.0000] that's clearly not a good plan [16:32:00.0000] Very interoperable, no doubt :) [16:34:00.0000] Would it be non-conforming to just leak memory and not do GC? [16:34:01.0000] so here's the problem [16:34:02.0000] You'd still be firing it every time you did GC, so it should be conforming [16:35:00.0000] this is for http://www.hixie.ch/specs/dom/messages/0.9 [16:35:01.0000] what should happen if i create an end point, set its message handler, and then forget about my end of it, having passed the other end to another browsing context? [16:35:02.0000] clearly it shouldn't be garbage collected [16:36:00.0000] as it can still receive messages and respond usefully [16:36:01.0000] but then what happens if my Window is closed? [16:36:02.0000] or navigated? [16:37:00.0000] Is the strong entangling really needed? [16:37:01.0000] should the other side get an unload event for the endpoint? [16:37:02.0000] how do you mean? [16:37:03.0000] Instead of just having each endpoint have an "other end" property referring to the other endpoint [16:38:00.0000] i don't understand [16:39:00.0000] Um. Well, could you treat it the same way any old DOM reference in a different window? [16:40:00.0000] With window references you have .closed, could something similar work for endpoints? [16:40:01.0000] i can set .active to false [16:44:00.0000] The problematic event was something like "other endpoint is gone"? [16:44:01.0000] i think i've worked out how to spec it for now [16:44:02.0000] we'll see if it handles Workers too in a second [16:45:00.0000] What's the reason for cloning endpoints sent via postMessage, by the way? [16:45:01.0000] cross-origin concerns [16:47:00.0000] for example, it would be bad if you could do foreignEndPoint.prototype.prototype.toString = function () { my evil function } [16:47:01.0000] or whatever [16:48:00.0000] So the create function is usually making one just to throw it away later [16:48:01.0000] yep [16:49:00.0000] I feel like it should be possible to do it the other way around, creating just one endpoint and having the semantics of send be "The endpoint received should be a new one entangled with this here local endpoint I have" [16:49:01.0000] what if you do it twice? [16:50:00.0000] also, you might want to pass the same end point down several pipes [16:50:01.0000] which you couldn't do the way you describe it [16:51:00.0000] Aren't those two things mutually exclusive? [16:51:01.0000] hm? [16:51:02.0000] If you want to pass it down several pipes, you are doing it twice (or more) [16:52:00.0000] say you have established pipes A1<->A2 and B1<->B2, and a newly created pipe C1<->C2. [16:52:01.0000] right now you can pass C2 down from A1 to A2, and then from there it can go from B1 to B2 [16:52:02.0000] but your proposal can't do that, as i understand it [16:53:00.0000] however, your proposal would allow sending "C1's other point" down A1 to A2 creating C2, and then sending "C1's other point" again down A1 to A2, creating C2'? [16:53:01.0000] so now you have two C2s [16:54:00.0000] which is confusing [16:54:01.0000] and causes all kinds of problems [16:54:02.0000] Let me work this out. So you pass C2 from A1 to A2, which gives A2 a C3 connected to C1. Then you pass C3 from B1 to B2, giving B2 a C4 which is connected to C1. In the process you've invalidated C2 and C3. [16:55:00.0000] yah [16:55:01.0000] (i'd prefer the notation C2, C2', C2'', C2''', but sure :-) ) [16:56:00.0000] Yeah, I see my idea wouldn't support chaining [16:57:00.0000] (Unless the endpoint included a pointer to its twin) [16:57:01.0000] And that's what the cloning was supposed to stop [16:58:00.0000] yeha [16:58:01.0000] yeah, even [16:59:00.0000] So I suppose I'd be writing function setupLink(w) { p = new Pair(); w.postMessage("New pair",p.two); return p.one; } or somesuch then [17:19:00.0000] hmmmmm [17:19:01.0000] workers have to die if no browsing contexts refer to them [17:19:02.0000] not just if nobody refers to them [17:19:03.0000] otherwise you could have a worker refer to another and vice versa [17:19:04.0000] and they'd keep each other alive [17:33:00.0000] I don't know how to address this comment for selectors api http://lists.w3.org/Archives/Public/public-webapi/2008Feb/0000.html [17:34:00.0000] hm, i wrote that text [17:34:01.0000] or suggested it originally [17:34:02.0000] let's see [17:34:03.0000] I can't figure out how to rephrase it to make it clearer [17:34:04.0000] yeah [17:34:05.0000] User agents should ensure that they do not crash or behave erratically when facing... [17:35:00.0000] that might work [17:35:01.0000] ... For example, a user agent could apply a timeout to the NSResolver callback. [17:35:02.0000] but then I'm not totally convinced of the value of having that in there at all, since the same applies to all scripts [17:36:00.0000] well, the difference here is that the script is driving a UA algorithm [17:36:01.0000] i could easily imagine implementations of the algorithm where returning inconsistent results would cause a crash [17:36:02.0000] this is just saying, "beware" [17:36:03.0000] which is all security sections ever say, really [17:36:04.0000] ok [17:37:00.0000] but i can easily imagine someone not thinking about the fact that the resolver might hang, and accidentally not putting the usual scripting timeout constraints around that api [17:37:01.0000] same with changing the dom [17:44:00.0000] http://www.w3.org/2001/tag/doc/passwordsInTheClear-52 [17:45:00.0000] "User agents SHOULD use password masking when passwords are displayed in an HTML form" [17:45:01.0000] i'm so glad the TAG is addressing critical unsolved problems [17:45:02.0000] and not, say, problems from 1993 [17:53:00.0000] http://www.hixie.ch/specs/dom/workers/0.9 is finally ready for broader review [18:18:00.0000] the recursive definition of "acceptable" is a bit dogy [18:19:00.0000] a cycle of WorkerWindows talking to each other but disconnected from all real Windows could be deemed "acceptable", and could also be deemed "not acceptable" [18:21:00.0000] you also need to say something about what's in scope for these worker threads [18:22:00.0000] oh I see you've got a one-line comment covering that :-) [18:27:00.0000] is createWorker*() designed to make it possible to run JS as a separate thread in the background? [18:27:01.0000] yes [18:27:02.0000] cool [18:27:03.0000] a la Gears [19:08:00.0000] roc: yeah, i thought of that when i defined it (the 'acceptable' thing). i was hoping it would go unnoticed. [19:08:01.0000] fat chance of that, apparently :-) [19:08:02.0000] roc: any idea how to better phrase it? [19:09:00.0000] something like [19:09:01.0000] "Acceptability is defined inductively by the following rules:" [19:09:02.0000] (these two proposals are very much early early drafts, by the way, hey'll obviously be defined more rigorously if they ever make a real spec) [19:10:00.0000] roc: is that magic phrasing that makes it better? awesome [19:10:01.0000] /me edits [19:10:02.0000] no [19:10:03.0000] "1. " [19:10:04.0000] oh, there's more, ok [19:10:05.0000] /me waits [19:10:06.0000] "2. " [19:10:07.0000] but yeah, "induction" is actually a magic word here :-) [19:11:00.0000] you might find it useful to take a class or read a book on methods for formal specification [19:12:00.0000] a lot of it's less useful than its practitioners wish [19:12:01.0000] but you might pick up some useful stuff anyway [19:12:02.0000] yeah, i try to read up on stuff like that occasionally [19:12:03.0000] do you have any suggestions? [19:12:04.0000] there's a lot of... less than readable material [19:12:05.0000] yeah [19:12:06.0000] that's either way over my head, or bogus [19:12:07.0000] i can't tell [19:14:00.0000] I'll ask my old PhD advisor(s), they're into this sort of thing [19:15:00.0000] cool [19:15:01.0000] that would be awesome [19:16:00.0000] BTW another way of phrasing those acceptability constraints would be to specify that you want the least solution to the constraints, i.e. the solution that makes the minimum number of workers acceptable [19:17:00.0000] true [19:17:01.0000] ok, i've tried taking your structure and applied it to the draft [19:17:02.0000] http://hixie.ch/specs/dom/workers/0.9 [19:17:03.0000] that's quite a popular approach, although sometimes defining "minimal" is itself quite tricky [19:18:00.0000] looks good [19:18:01.0000] cool [19:18:02.0000] thanks [19:28:00.0000] Hixie: weren't you originally an English major? How much mathematics have you done? [19:28:01.0000] physics [19:28:02.0000] ah ok cool [19:29:00.0000] so you probably learned lots of calculus and geometry, which is useless :-) [19:29:01.0000] pretty much [19:29:02.0000] and all the stats i learnt, which would have been useful, i forgot [19:30:00.0000] the scientific method is the main thing i got from my degree which i've applied to the spec work [03:04:00.0000] passwordsInTheClear-52 lol [03:06:00.0000] I wonder if I should point out a small mistake or not [03:12:00.0000] annevk, what's the mistake? [03:13:00.0000] Is this mail basically asking what string can be used for hasFeature() to test for selectors api? http://lists.w3.org/Archives/Public/public-webapi/2008Feb/0035.html [03:13:01.0000] I think so [03:13:02.0000] I don't see a compelling reason to define one [03:14:00.0000] Me neither [03:14:01.0000] if(document.whateverthenamewas) [03:14:02.0000] exactly [03:14:03.0000] since hasFeature virtually never gets used int he wild and that alternative already works [03:14:04.0000] and since hasFeature must die [03:15:00.0000] :-) [03:16:00.0000] /me stabs hasFeature [03:19:00.0000] Lachy, as for -52, the second example they give in the HTML section is something the UA can't know and the Web author does and therefore the Web author doesn't use a password field at all and therefore the example is wrong [03:21:00.0000] /me wonders how something like target="_blank-but-with-more-security-separation" would work, since JS code will have to determine whether the UA has that feature before deciding to use it (rather than falling back to window.open tricks) and that seems impossible to determine unless there's some API call that returns true when that feature is supported [03:22:00.0000] the same is true for rel=noreferrer [03:22:01.0000] oh, maybe rel=noreferrer is the solution? [03:23:00.0000] /me forgot [03:24:00.0000] annevk, I only see 1 HTML example, which appears to be missing the name attribute [03:24:01.0000] in section 3 [03:24:02.0000] I guess in practice people would just use 'if (navigator.userAgent.indexOf('Firefox/4.0.) == -1) { /* use slow ugly fallback mechanism */ }' if they're not given any better way of determining supportedness [03:25:00.0000] Lachy, the text that provides exceptions in terms of examples to the SHOULD-level requirement is what I was talking about [03:25:01.0000] Philip`, yeah, transition periods are awkward now and then [03:27:00.0000] We're always in a transition period - nothing is going to settle down and stabilise until it's dead [03:30:00.0000] individual features have shorter transition periods than that though [03:30:01.0000] I agree that the Web as a whole has a slightly different model :) [03:37:00.0000] /me wonders if by 2010 he'll be able to use transparent PNGs without style="-filter:progid:DXImageTransform.Microsoft.AlphaImageLoader(...)" [03:40:00.0000] /me has used transparent PNGs without that kind of stuff since circa 2000 [03:41:00.0000] /me also needs the transparency to actually be transparent, and needs to work for a sufficiently large proportion of users [05:04:00.0000] I just have 3 more issues to resolve in selectors api, and I'm done. [05:05:00.0000] did you make a simple plain text issues list already? [05:05:01.0000] I'm not sure how to deal with these 2 regarding the IDL (see points 2 and 3 in this mail) http://lists.w3.org/Archives/Public/public-webapi/2007Dec/0072.html [05:05:02.0000] point to DOM Bindings and say that we're going to use that [05:06:00.0000] not yet, I've just been tagging the mails in my mail archive. I will write out a list of issues from that later [05:06:01.0000] k [05:06:02.0000] in due course the Web API WG will use http://www.w3.org/TR/DOM-Bindings/ for its deliverables once that reaches a slightly more stable state [05:07:00.0000] so currently the IDL syntax is more of an informative nature, really [05:07:01.0000] the 3rd issue is the one about using XHTML syntax instead of HTML syntax in examples http://lists.w3.org/Archives/Public/public-webapi/2008Jan/0004.html [05:07:02.0000] and we do plan to fix that before going to PR [05:07:03.0000] changing examples always introduces bugs is my experience [05:08:00.0000] I thought about using DOM Bindings, but wasn't sure, since it's not finished yet [05:08:01.0000] I'll check in my current revision now [05:08:02.0000] we'll do DOM Bindings before CR or PR, whenever DOM Bindings itself is in PR/REC [05:09:00.0000] (well, that's my idea anyway) [05:13:00.0000] When I buy postage stamps, I want them to have the Queen's head on them, not the word "LOVE" in large pink letters :-( [05:17:00.0000] I wonder why they have LOVE in large pink letters [05:20:00.0000] ann latest version checked in http://dev.w3.org/2006/webapi/selectors-api/ [05:20:01.0000] s/ann/annevk/ [05:30:00.0000] /me sends off all responses to the last call comments for selectors api to public-webapi [05:35:00.0000] I'm not really convinced with the exceptions story, still [05:35:01.0000] even thoug I added them [05:35:02.0000] using querySelector to test selector support seems wrong [05:38:00.0000] well, send mail about it to public-webapi and see what others think about the idea of removing exceptions [05:38:01.0000] i remember having done that before [05:42:00.0000] you only sent me mail privately [05:42:01.0000] I don't recall seeing any discussion about removing exceptions on the mailing list [05:43:00.0000] here's that feedback I got from mjs about exceptions http://krijnhoetmer.nl/irc-logs/whatwg/20080208#l-17 [05:48:00.0000] i'm pretty sure it was on the list too and bjoern didn't like it for similar reasons as maciej or something [05:52:00.0000] if you did, I can't find it in pubilc-webapi [07:26:00.0000] Hmm, Firefox doesn't like setAttribute(':', ...) [07:27:00.0000] I think jwalden might have a patch for that? [07:27:01.0000] oh, no, I was thinking of bug 411103 [07:29:00.0000] /me tries filing https://bugzilla.mozilla.org/show_bug.cgi?id=417491 [07:33:00.0000] data:text/html,%3Cbody%20:=""%20onload=%22try%20{%20document.body.setAttribute(':',%20'');%20alert('OK')%20}%20catch(e)%20{%20alert(e)%20}%22%3E works, though [07:33:01.0000] there's a bug on this already, I think [07:34:00.0000] https://bugzilla.mozilla.org/show_bug.cgi?id=390816 [07:35:00.0000] gavin_: That looks different - ';' should be rejected because it's not a valid XML Name [07:36:00.0000] Philip`: ":" is not a proper XMLNS name, though [07:37:00.0000] (although Gecko has a dual XML 1.0 / XMLNS 1.0 nature in some things, IIRC) [07:37:01.0000] hsivonen: DOM just says it should be an XML name - the namespace things only affect setAttributeNS etc [07:39:00.0000] hmm [07:39:01.0000] /me gives up trying to work out what DOMs can be safely constructed, and just wraps stuff in exception handlers [07:39:02.0000] i think it makes sense to require XMLNS Name [07:40:00.0000] so createElement can just become a shorthand for createElementNS(htmlns, name) [07:41:00.0000] /me agrees with annevk [07:41:01.0000] /me tries to understand MOZ_XMLCheckQName [07:41:02.0000] on a second though, that won't work for XHR [07:41:03.0000] so, no [07:42:00.0000] since IE XHR is DOM Level 1, stuff like the Flickr API have to be XML 1.0 without XMLNS [07:42:01.0000] yay for XMLNS [07:43:00.0000] i don't understand [07:43:01.0000] xhr requires a conforming xmlns parser [07:44:00.0000] IE6 doesn't expose a namespace-aware API for XHR-loaded docs, IIRC [07:45:00.0000] and Safari as shipped with Panther missealized namespaced XML [07:45:01.0000] so RESTful XML APIs that want to be XHR-compatible have to live in the namespaceless XML 1.0 (< 5th ed.) world [07:45:02.0000] /me gives up trying to understand, uses a debugger instead [07:46:00.0000] isserialized [07:46:01.0000] misserialized [07:46:02.0000] typo++ [07:46:03.0000] well namespaceless is fine [07:47:00.0000] i'm just saying that they have to compatible with XMLNS for things to work [07:48:00.0000] sure [07:48:01.0000] http://philip.html5.org/tools/parser/?%3C!DOCTYPE%20HTML%3E%0A%3Cx%3Ax%20x%3Ax%3E%3Cx%3A%3Ax%20x%3A%3Ax%3E%3Cx'x%20x'x%3E%3Cfoo%20%3Ax%3E [07:48:02.0000] Opera, Firefox and Safari all give different output :-( [07:48:03.0000] but at least Opera 9.2 and 9.5 are the same, and Firefox 2 and 3 are the same [08:12:00.0000] Philip`, heh, cool stuff [08:17:00.0000] weinig, jwalden (if you read this), any feedback on adding reply() ? [08:28:00.0000] webben: Aaron's reply is interesting [08:29:00.0000] annevk: ^^^ [08:29:01.0000] webben: sorry, misdirected that message [08:30:00.0000] yeah, though then you want a callback for reply() too for longer conversations... [08:31:00.0000] hmm [08:33:00.0000] annevk: true [08:33:01.0000] annevk: you could end up with some very long callback chains on either end [08:33:02.0000] annevk: (well, I guess both ends, since the chains would have to be within one callback of each other in length) [08:34:00.0000] yeah, seems icky [08:35:00.0000] maybe postMessage should return a uuid and that uuid is exposed on the event [08:35:01.0000] and when you invoke reply it automatically uses the same uuid [08:36:00.0000] annevk: I guess that would be a little easier than what Gears does [08:36:01.0000] annevk: since with Gears you have to embed the ID into your message string [08:36:02.0000] yeah, with postMessage() you'd have to do that too now [08:37:00.0000] annevk: right [08:48:00.0000] ok, posted that idea [08:48:01.0000] i have a feeling that hixie's endpoints stuff does similar tricks, we'll see [08:49:00.0000] but this proposal is not that more complex than what we have now, so it should be easier to have browsers updated in time with everyone's nearing release dates [08:49:01.0000] annevk: "endpoints"? [08:50:00.0000] http://hixie.ch/specs/dom/messages/0.9 [08:50:01.0000] i haven't carefully checked it yet after asking hixie for more rationale [08:50:02.0000] /me reads [09:03:00.0000] hm, I think there's a typo in the example on that page [09:03:01.0000] gadgetWindow.endPoint.postMessage('get list-of-names', event.endPoint); [09:03:02.0000] seems like it should just be [09:04:00.0000] gadgetWindow.postMessage('get list-of-names', event.endPoint); [09:05:00.0000] Hixie: ^^^ [10:04:00.0000] oops. fixed. [10:36:00.0000] Ooh, a tokeniser change - how exciting [10:39:00.0000] /me updates his four tokeniser implementations [10:40:00.0000] hah [10:41:00.0000] the graph of time-from-message-being-sent against probability of message being replied that day has a spike at t = 0 days [10:43:00.0000] Hixie: do you have the graph online? [10:46:00.0000] P| [10:46:01.0000] | [10:46:02.0000] |_ [10:46:03.0000] | | [10:46:04.0000] | | ____ ? [10:46:05.0000] | |___...---''' [10:46:06.0000] | [10:46:07.0000] -+-------------------->t [10:47:00.0000] the start of it -- a few months -- looks like that [10:47:01.0000] dunno what the whole thing looks like [10:47:02.0000] Hixie: thanks [10:52:00.0000] It's kind of odd how the Ruby html5lib claims to pass all the tokeniser tests, despite it not actually passing [10:53:00.0000] Aha, it does "return expected == received" instead of "assert_equal expected, received" [10:53:01.0000] hmm. I wonder what the perf implications of query strings are with the new Access-Control-Policy-Path thing [10:53:02.0000] /me rereads the spec [10:54:00.0000] hmm. looks good [10:55:00.0000] now awaiting for an access-control-enabled browser, then [10:55:01.0000] for validation mashups [11:07:00.0000] So, I think the Python and Ruby and OCaml and C++ and JS and Perl tokenisers now match the spec again [11:09:00.0000] http://philip.html5.org/tools/parser/?%3C!DOCTYPE%20HTML%20%225%3E2%22%3E%0A... [11:09:01.0000] Hmm, that totally doesn't work right in Opera 9.2 [11:17:00.0000] Philip`: you want http://philip.html5.org/tools/parser/?%3C!doctype%20html%20public%20%22%3E%20X [11:18:00.0000] heh, I love how things are only bugs if one actually claims conformance :-) http://lists.w3.org/Archives/Public/public-webapi/2008Feb/0071.html [11:18:01.0000] Is RIM a browser vendor these days? [11:18:02.0000] zcorpan_: Argh, good point [11:21:00.0000] Lachy: I assume that means "claim" in the "returns true from hasFeature" sense, not in the "marketing material has it as a bullet pointed feature"; i.e. it would be alright for someone to have an incomplete/buggy implementation if hasFeature says it's not properly supported and shouldn't be used [11:21:01.0000] which seems more reasonable than how I initially read it :-) [11:22:00.0000] so if hasFeature returns false for everything, the UA is bug-free? [11:22:01.0000] zcorpan_, yes. [11:22:02.0000] awsome [11:22:03.0000] /me goes to file a bug for opera about hasFeature returning true for somethings [11:23:00.0000] /me wonders whether there is a feature for hasFeature working... [11:23:01.0000] Core 2.0 or something? [11:23:02.0000] So if hasFeature for that returns false, doens't that mean that other return values are irrelevant? :-) [11:23:03.0000] SadEagle, I think it would be covered under the "DOM 2.0" features string (or whatever it is) [11:24:00.0000] hasFeature("hasFeature", "1.0") [11:24:01.0000] false [11:24:02.0000] :) [11:25:00.0000] If you return false for hasFeature then you will get the sort-of-gracefully degraded fallback path that authors have conscientiously provided for user agents that don't have the feature, and your users will hate you because they get a degraded experience, and so market forces will compel you to implement the feature perfectly so you can return true from hasFeature [11:28:00.0000] everything should just be implemented perfectly in every browser on the first attempt, and then hasFeature can be changed from a method to a single property that just equals true. [11:30:00.0000] Philip`: might be interesting to see how many websites actually use hasFeature [11:32:00.0000] http://google.com/codesearch?q=show:mUNqRgxPl_s:oKMDwuYvdP4:CNlRzrngUHA&cs_p=http://www.devlib.org/apache/struts/source/struts-2.0.6-src.zip&cs_f=struts-2.0.6/src/core/src/main/resources/org/apache/struts2/static/dojo/dojo.js.uncompressed.js#l1157 [11:32:01.0000] There's one [11:33:00.0000] http://google.com/codesearch?q=show:BNeIttp5oHc:5g2Z17w4TXQ:BXCuyDas2KE&cs_p=http://openrico.org/svn/openrico/trunk&cs_f=openrico/public/dist/1.1.2/rico.js#l1337 [11:34:00.0000] http://google.com/codesearch?q=show:UWuK4cTDdZc:A3Gk8FzZWSM:L1InoVp7Omk&cs_p=http://gentoo.osuosl.org/distfiles/Plone-2.1.4.tar.gz&cs_f=Plone-2.1.4/kupu/common/sarissa_ieemu_xpath.js#l26 [11:34:01.0000] It seems easy to find quite a few [11:36:00.0000] heh, #1 follows it up with a UA sniff. #2 uses IE-specific path as a fallback, and will fail in konqueror, since it tests for "CSS", "2.0" --- odd thing to test if you want -events- support. #3 actually looks right :-) [12:32:00.0000] "Making an implementation bug-free is outside the scope of the specification." -- http://www.w3.org/mid/24BE9C481C439C48831EB9E52639B9B60A8D4842⊙Xrn [12:32:01.0000] hmm, perhaps i have misunderstood the purpose of having specs [12:34:00.0000] it's to say what 'bug-free' means :-) [12:35:00.0000] i think its time for me to finetune my ignore flag again [12:36:00.0000] I'm bug free. [12:53:00.0000] zcorpan_, technically that statement is correct, depending on how you read it. It's the implementor's responsibility to actually write the bug free code, not the specification. But it is the specification's responsibilty to ensure that things are defined well enough to be implemented in bug free ways [13:42:00.0000] RFC3986 doesn't reference RFC2119! [13:42:01.0000] /me is amazed [13:45:00.0000] they don't capitalize SHOULD/MAY/etc. [13:45:01.0000] http://lists.w3.org/Archives/Public/public-webapi/2008Feb/0076.html [13:45:02.0000] what on earth [13:46:00.0000] Sent from Yahoo! Mail - a smarter inbox [13:51:00.0000] Lachy, one other thing to consider btw is to call StaticNodeList NodeList again and mention specifically that DOM Level 3 Core (and maybe previous versions) contains a bug as it calls it a live list where it just defines an interface [13:52:00.0000] isn't that what i said at the time :-P [13:53:00.0000] i was convinced, but some other people were not [13:53:01.0000] probably easier to get it through now, especially with webkit already doing it that way! [13:54:00.0000] man, these SMIL people are pushy [13:55:00.0000] hm? [13:56:00.0000] not really [13:56:01.0000] just getting quite a bit of e-mail about this timesheets draft [13:57:00.0000] timesheets? [13:57:01.0000] wtf is that [13:57:02.0000] That widely implemented draft, obviously. [13:58:00.0000] That would be easier [13:58:01.0000] I had assumed that window.DocumentSelector and window.ElementSelector would be exposed, since the interfaces for other DOM2 APIs are exposed [13:58:02.0000] /me assumes timesheets is for doing things that were common on 90s era CD-ROMS [14:00:00.0000] aargh! What is wrong with Sergey's email client? Why is it putting one word per line?! [14:00:01.0000] one of the spec editors demoed me timesheets [14:00:02.0000] I think DocumentSelector and ElementSelector should not be exposed on Window [14:00:03.0000] the demoed use case was doing S5 without the JS [14:00:04.0000] /me bravely presses send [14:00:05.0000] they are redundant [14:00:06.0000] hopefully DOM Bindings provides a way to say that [14:01:00.0000] I quite liked the Apple CSS animation stuff for transitions and the like [14:01:01.0000] yeah, we need that [14:01:02.0000] Lachy: are DocumentTraversal, DocumentViews, ElementTarget, DocumentStyle, etc. exposed? [14:01:03.0000] fits what already is there much better than SMIL [14:01:04.0000] but the $%# CSS WG declared it out of charter [14:02:00.0000] I don't claim my assumptions about the purpose of timesheets are at all accurate [14:02:01.0000] SadEagle, Firefox exposes DocumentTraversal and DocumentStyle [14:03:00.0000] Odd. But then the entire thing is a mozillaism everyone else had to emulate, making it a defacto standard :-) [14:03:01.0000] SadEagle, EventTarget should be exposed, not sure what ElementTarget is [14:03:02.0000] It's a very problematic thing to do, though. [14:03:03.0000] I agree that exposing DocumentStyle and such makes no sense [14:03:04.0000] annevk: that's what I meant... And I guess EventTarget makes sense, since it can be used as a base class. [14:03:05.0000] Well, what should DocumentStyle.prototype point to? [14:04:00.0000] FF exposes EventTarget too. WebKit doesn't seem to expose any of them [14:04:01.0000] probably maps to Document.prototype [14:04:02.0000] I don't really like what Firefox does with interfaces at all [14:04:03.0000] why not? [14:05:00.0000] there's now some Web content that relies on XMLDocument and HTMLSpanElement, HTMLUnknownElement and crap like that [14:05:01.0000] It makes sense for the ones that define constants, like Node.ELEMENT_NODE [14:05:02.0000] well, that stuff makes sense [14:05:03.0000] that stuff is also far more interoperable than what we're discussing here [14:06:00.0000] good lord, the html wg is considering even more mailing lists [14:07:00.0000] Hixie, what more mailing lists are being considered? [14:07:01.0000] see #html-wg [14:07:02.0000] annevk: those are the interfaces they have aren't in the DOM, right? [14:10:00.0000] yeah [14:17:00.0000] /me ponders the idea of adding a hasFeature string to selectors api that implementors are not required to support :-) [14:18:00.0000] hasFeature('Selectors.hasFeature') [14:19:00.0000] good to know that the chair of the html working group doesn't read all the mail to the html working group [14:22:00.0000] /me doesn't find that too encouraging [14:23:00.0000] /me wonders how many do [14:23:01.0000] Chairs of the WG or participants in general? [14:23:02.0000] everyone in the WG [14:23:03.0000] I skim each e-mail, though I delete some right away based on subject/sender [14:24:00.0000] (that's not specific to the HTML WG though) [14:24:01.0000] i read everything, and save anything that looks useful to the folders for further study later [14:25:00.0000] i read public-html (and whatwg, and public-appformats and -webapi) before reading anything from the groups i'm not as actively in, like css or tag or http [14:25:01.0000] I ignore anything to do with vs as a matter of course. And occasionally I never get round to reading some messages. [14:25:02.0000] I skim most things, though there are some things which I know so little about I just ignore from the subject [14:26:00.0000] I tend to look at all the words in everything, but I'm not sure if it counts as reading [14:26:01.0000] Mostly it's just a way of getting rid of the 'unread messages' marker [14:27:00.0000] yeah well some e-mails certainly don't get a very thorough reading [14:27:01.0000] especially those written in reply to e-mails that themselves didn't seem to get a thorough reading [14:27:02.0000] and i rarely spend more time reading an e-mail than the other person spent writing it [14:28:00.0000] (or thinking about it) [14:28:01.0000] Hixie: I guess that means 10 seconds for most of my emails :) [14:28:02.0000] (sadly) [14:30:00.0000] :-) [14:33:00.0000] http://weblogs.mozillazine.org/doron/archives/2008/02/exe_tld_coming_soon.html :o [14:33:01.0000] /me squints [14:33:02.0000] that's not a new issue, though, see .pl :-) [14:33:03.0000] silly Poland, thinking they deserve a TLD! [14:34:00.0000] Or perhaps: Why's that relevant? I mean, who uses perl nowadays? [14:35:00.0000] SadEagle, heh, true [14:35:01.0000] /me is too tired to be truly witty [14:35:02.0000] SadEagle, well, actually, .pl is not handled by the browser [14:35:03.0000] i'm mildly amused at how the w3c staff (except for mike) aren't actually an active part of the html5 community, despite them making process decision on the htmlwg [14:36:00.0000] i don't really like it [14:37:00.0000] annevk: it could be though, viewed as text... I am not sure where this distinction matters in mozilla, though (it showed up in KDE, but that was outside a web browser) [14:38:00.0000] i wonder where browsers do extension sniffing though [14:38:01.0000] HTML5 contains no such thing! [14:38:02.0000] there's also .au [14:38:03.0000] browsers do extension sniffing in the places where html5 explicitly says not to do extension sniffing [14:39:00.0000] there was a famous Gecko bug that only happened on files downloaded from Australia [14:39:01.0000] that took a while to figure out [14:39:02.0000] hah [14:42:00.0000] /me waves g'nite [14:44:00.0000] 'course if .com is no problem, why would .exe be a problem [14:46:00.0000] http://eicar.com/download/eicar.com [14:48:00.0000] Dashiva, you registered? [15:09:00.0000] annevk: Where? [15:11:00.0000] freenode, nm though, you're on w3.org too :) [15:39:00.0000] Philip`: the reason one is a list and one is as stack (despite neither being either) is exactly because it makes them easier to distinguish 2008-02-15 [16:04:00.0000] since the htmlwg is having a meeting, let's have one too. [16:04:01.0000] anyone have any outstanding action items? [16:04:02.0000] oh look, we've never assigned any. [16:04:03.0000] ok, meeting over. [16:04:04.0000] Please come run my company's next monthly meeting. [16:05:00.0000] my solution to meetings is very simple [16:05:01.0000] i refuse to attend any meeting that recurrs [16:05:02.0000] i welcome anyone to book me to any meeting they like, but if it recurrs, i won't go. [16:06:00.0000] (i also generally refuse action items that i can't do during the meeting) [16:06:01.0000] I think I'll put that on a t-shirt. "i refuse to attend any meeting that recurrs" [16:07:00.0000] Hixie: a novel approach, but you have to grant a bit of beaurocracy [16:07:01.0000] why? [16:07:02.0000] we've managed the whatwg pretty well without any [16:07:03.0000] billmason: fix the typo first (recurrs -> recurs) [16:07:04.0000] i was wondering about that [16:08:00.0000] i have 2-3 recurring meetings every week, each about an hour [16:08:01.0000] i'm so sorry [16:08:02.0000] Hixie: just because the w3 is actually a beaurocracy, and to be a valid one, it needs recurring meetings and action item trackers and all that [16:08:03.0000] hehe [16:09:00.0000] takkaria: so you're saying i have to allow the w3 to be a valid bureaucracy? [16:09:01.0000] i'd rather not :-) [16:09:02.0000] I'm not even saying you should go to the meetings [16:10:00.0000] I'm just saying that getting html5 the w3c patent review and all that outweighs the silly meetings [16:11:00.0000] i agree that the patent policy is a useful thing [16:11:01.0000] the jury's still out on whether the cost of working through the w3c outweighs the benefit of the patent policy [16:12:00.0000] I have annoying daily project meetings at work, lasting up to half an hour. I hate it [16:12:01.0000] DAILY?! [16:12:02.0000] yes [16:12:03.0000] jeez [16:13:00.0000] I used to be on a project like that. It was one of the reasons I got off of it.... [16:14:00.0000] I'm attempting to use html5lib to process a vml document. vml doesn't support css very well, so if I want to use a common stylesheet to apply styles to both svg and vml documents representing the same thing, I'll need to parse the stylesheet, find all the elements seeming to match any rules, and then translate the properties of applying rules into a dom transformation that will apply the desired styles [16:15:00.0000] i have three recuring meetings total, one every two weeks which is pretty much the only time i meet my team mates and find out wtf they're doing, one every week which is the campus-wide beer-and-news-and-questions-for-the-management fest, and one semi-annual (or quarterly? i forget) team week, of which i only attend half. [16:15:01.0000] bewest: so you want a css parser? [16:16:00.0000] yes, among other things [16:16:01.0000] I think the cssutils python package is probably usable [16:16:02.0000] I think I need to use the etree builder with html5lib [16:17:00.0000] when I tried this, etree complains about the doctype before the treebuilder has finished instanciating [16:17:01.0000] Hixie: what kind of team are you on? are you all web standards-related people or is it just a "misc" team? [16:17:02.0000] (before parsing begins) [16:18:00.0000] takkaria: i'm in the open source program office. same team that does google summer of code, googlecode.com, etc [16:18:01.0000] ah, I see [16:18:02.0000] we're kind of an "odds and ends" team [16:18:03.0000] but mostly open-source related [16:19:00.0000] can anyone suggest alternative strategies I should try to acheive consistent styling across vml and svg? [16:20:00.0000] does html5lib expect certain versions of etree? [16:23:00.0000] http://dpaste.com/35244/ is the traceback of http://dpaste.com/35245/ [16:25:00.0000] my understanding is that I should be able to get etree to turn a css selector into an xpath [16:26:00.0000] hmmm maybe I should check the treebuilder's source for compatibility hints [16:31:00.0000] Hixie: can you tell your teammates that the svn browser on googlecode breaks the back button? [16:31:01.0000] it does?L [16:32:00.0000] aaah [16:32:01.0000] they changed the interface [16:32:02.0000] wtf [16:33:00.0000] indeed [16:33:01.0000] wow yeah, that's annoying [16:34:00.0000] very [16:34:01.0000] makes the interface much less efficient [16:34:02.0000] seems you can still get to the old interface [16:35:00.0000] /me doesn't see how [16:35:01.0000] go to a file [16:35:02.0000] choose "raw view" [16:35:03.0000] and then navigate from that uri [16:35:04.0000] aha [16:37:00.0000] apparently it's a known bug [16:37:01.0000] if it ain't broke... ;-) [16:38:00.0000] well, the old ui didn't show revisions [16:38:01.0000] oh [16:38:02.0000] which apparently was a huge issue [16:38:03.0000] yeah, it's nice to have revisions [17:46:00.0000] ps. i hate utf-16 [17:47:00.0000] /me grumbles at anne for making the utf-16 disease creep into svg [17:58:00.0000] Hixie, what's exactly is the UTF-16 disease? [18:03:00.0000] http://weblogs.mozillazine.org/roc/archives/2008/01/string_theory.html [21:14:00.0000] /me ponders how to make pushState() work when you cross them when the document isn't active [21:14:01.0000] consider something like this: [21:17:00.0000] new browsing context, navigate to A. navigate to B. pushState B1, pushState B2. navigate to C. [21:17:01.0000] now go back twice, activating B1 and B2 so that the current entry is B1 [21:18:00.0000] now go back twice again, which puts you on A. [21:18:01.0000] now jump to the end, which puts you on C. [21:18:02.0000] now go back one. [21:19:00.0000] we need both B1 and B2 to fire, but... [21:21:00.0000] hmm, pushState() as defined doesn't work right at all [21:21:01.0000] well bummer. [21:34:00.0000] hmmmmmmmm [21:34:01.0000] so it seems we need an event to fire when you switch to an entry that isn't a state entry [21:34:02.0000] then everything is fine [01:05:00.0000] Hixie, hmm, I wasn't the one who added that test, I'm just fixing SVG to be consistent :p [01:46:00.0000] hsivonen: You might be interested in http://blog.mozilla.com/dmandelin/2008/02/14/wtf-16/ [01:48:00.0000] jgraham: thanks. Yeah, UTF-16 is overall a mistake. For some environments (like Java) it is a mistake not worth fixing, though. [01:49:00.0000] why? because Java is dead? [01:49:01.0000] jruderman: no, because having an undisturbed Java-wide consistent UTF is too useful [01:52:00.0000] I think roc is right about iterators [01:52:01.0000] in general, it is a mistake to let app code index into the string backing buffer [02:45:00.0000] annevk, yt? [02:46:00.0000] annevk, to address Boris' comment about supporting the same set of selectors in both CSS and Selectors API, should I make that a SHOULD or MUST requirement? [02:54:00.0000] hmm. Firefox 2 makes the whole GUI on Mac OS X 10.5.2 unusable [02:54:01.0000] for me at least. repeatedly [02:54:02.0000] no such problem with minefield [02:54:03.0000] working fine here [02:55:00.0000] (but admittedly, I don't use Fx much) [02:55:01.0000] I guess it is time to migrate my main profile to Firefox 3 [02:56:00.0000] annevk, I have added this to the spec. "If the user agent also supports some level of CSS, the implementation must support the same set of set of selectors that are supported by the CSS implementation." [02:56:01.0000] I'll check it into cvs later [04:00:00.0000] Lachy, looks good [04:15:00.0000] annevk: is the xml-dev thread about slimming XML on your XML5 radar? [04:16:00.0000] no [04:16:01.0000] sounds good though [04:20:00.0000] I got the SVG WG to use 16-bit units btw... [04:20:01.0000] nice [04:21:00.0000] http://lists.xml.org/archives/xml-dev/200802/msg00149.html [05:00:00.0000] have I got it right that postMessage() can be implemented as in the spec and that whatever extension will come will be on top of that? [05:01:00.0000] or is postMessage() going to be revised soonish, once again? [05:05:00.0000] current spec: void postMessage(in DOMString message, in DOMString origin); [05:05:01.0000] Hixie's proposal: void postMessage(in DOMString message, in EndPoint endPoint, in DOMString origin); [05:06:00.0000] true, but is that done through overloading or should only the latter be supported? [05:07:00.0000] i guess my real question is what Firefox 3 will ship so we can align with them [05:36:00.0000] http://lists.w3.org/Archives/Public/public-xml-core-wg/2008Feb/thread.html#msg26 [05:36:01.0000] public-xml-core-wg realizing how the Web functions [05:38:00.0000] Hmm, I get an infinite loop on
(whereas html5lib and validator.nu don't), and I can't tell that I'm wrong [05:40:00.0000] that's
, no? [05:41:00.0000]
i think [05:41:01.0000] yeah [05:42:00.0000] The
gets reprocessed in "in table" mode, which creates a
that gets processed in that mode, which generates implied end tags, and the current node is a so it generates a , which gets ignored because it's "in table" [05:42:01.0000] (and then it generates implied end tags for eternity) [05:42:02.0000] Act as if an end tag token with the tag name "table" had been seen, then, if that token wasn't ignored, reprocess the current token. [05:42:03.0000] Note: The fake end tag token here can only be ignored in the fragment case. [05:43:00.0000] It never gets as far as that "if" [05:43:01.0000] because the act-as-if- never terminates [05:44:00.0000] Philip`, only should be treated as-if [05:44:01.0000] not the other new tokens [05:45:00.0000] Hmm, that seems quite non-obvious [05:47:00.0000] I've interpreted it as being "old = mode; mode = InTable; process(
); if (mode wasn't changed while processing) mode = old" [05:54:00.0000] we do mode = newmode; process(...); mode = currentmode [05:54:01.0000] iirc [05:56:00.0000] Wouldn't that mean the
which gets generated and processed inside the process(...) would be done in 'newmode' ("in table") too? [05:57:00.0000] Seems to just do self.parser.phases["inTable"].processStartTag(name, attributes) [05:58:00.0000] yeah, sorry [06:00:00.0000] Philip`: IIRC, at least in v.nu parser old mode is only needed for moving from the endgame back to in body or in frameset [06:20:00.0000] ergh. the spec-gen is so badly documented. [07:04:00.0000] Lachy, why does the latest CVS version not use the Editor's draft template? [07:04:01.0000] can I quickly fix that, as well as the date? [07:04:02.0000] or are you editing? [07:20:00.0000] I've made some edits that still need to be checked in [07:21:00.0000] oh, ok [07:21:01.0000] crap, it should be using the editor's draft template. [07:21:02.0000] can you do that? [07:22:00.0000] Yes, it will get fixed with the next check in [07:23:00.0000] i meant checking in [07:25:00.0000] Hmm, Android SDK includes Ogg Vorbis [07:25:01.0000] i did not know [07:26:00.0000] all fixed [07:27:00.0000] Hmph, html5lib cheats with reprocessing and fostering [07:27:01.0000] like if you say then the implied doesn't go through the "in table" mode [07:27:02.0000] /me needs to learn how to use W3C CVS from my Mac, instead of having to do it from my home PC all the time [07:28:00.0000] you can only say cheats if we fail a test :) [07:28:01.0000] Philip`: surely the top of the stack is no longer IN_TABLE at that point? [07:28:02.0000] You fail a test that does
and expects a certain number of parse errors :-) [07:28:03.0000] Lachy, the date is still broken [07:28:04.0000] or do I misremember stuff? [07:28:05.0000] Lachy, you should omit the date parameter [07:29:00.0000] it has been a while since I have thought about the tree builder [07:29:01.0000] Philip`, it's ok to supress parse errors i believe :) [07:29:02.0000] hsivonen: Ah, yes, the shouldn't get fostered, but it should still go through "in table" and generate a parse error [07:30:00.0000] /me can't find an instance in html5lib where it would cause a different parse tree [07:31:00.0000] oops, I still had the date set in my .bat file. Fixed [07:33:00.0000] Lachy, did you see the e-mail about dumping StaticNodeList? [07:33:01.0000] the one from you in response to me? [07:34:00.0000] http://lists.w3.org/Archives/Public/public-webapi/2008Feb/0080.html ? [07:34:01.0000] yeah [07:34:02.0000] I will do it [07:34:03.0000] jgraham__: the commonality thread you replied to reminds me of http://www.nytimes.com/2008/01/20/magazine/20wwln-freak-t.html?_r=4&ref=magazine&oref=slogin&oref=slogin&oref=slogin&oref=slogin [07:35:00.0000] though sending that link to the list would probably be seen as a flame [07:36:00.0000] it's a rather complicated change, since I need to change/remove all occurrences of StaticNodeList, define the NodeList to be returned as not live and make sure no errors are introduced in the process [07:37:00.0000] true [07:37:01.0000] I'll give it a go [07:45:00.0000] annevk, any suggested text for how to define the node list as static? [07:48:00.0000] return the list of elements as a static object implementing the NodeList interface [07:48:01.0000] something in that direction [07:49:00.0000] or maybe "static NodeList object" [07:54:00.0000] /me ends up changing his 'fostering' flag to be an integer, so he can handle nested fostering without cheating [07:55:00.0000] http://philip.html5.org/tools/parser/?%3Ctable%3E%3Ctr%3E%3Ctable%3E - now with slightly fewer infinite loops, hooray [07:58:00.0000] annevk, "The NodeList object returned by the querySelectorAll() methods must not be live. Subsequent changes to the structure of the underlying document must not be reflected in the NodeList object. The object must instead contain a list of matching Element nodes that were in the document at the time the list was created." [08:00:00.0000] s/must not be live/must be static/ [08:01:00.0000] or s/must not be live/must be static (not live)/ [08:01:01.0000] the other statements can be phrased as statements of facts instead of conformance criteria [08:03:00.0000] "must be static (not live)" works, where _live_ is a link to the defintion of live lists in DOM3 [08:03:01.0000] i think it's universally understood what it means, but you could do that [08:04:00.0000] yeah, I just wanted to make it absolutely clear [08:06:00.0000] http://www.w3.org/mid/op.t2kr1qnf64w2qv⊙aooc - internal server error? [08:06:01.0000] wfm [08:06:02.0000] http://lists.w3.org/Archives/Public/public-forms-tf/2007Nov/0000.html [08:07:00.0000] Oh, it worked on the third attempt [08:22:00.0000] oops... why did I confuse vorbis with theora... [08:22:01.0000] annevk, changes have been committed [08:22:02.0000] let me know if I buggered anything up [08:23:00.0000] /me goes to friday beer [10:57:00.0000] /me wonders how the spec-gen deals with broken stuff [10:59:00.0000] annevk: postMessage() is reasonable stable as currently in the spec [10:59:01.0000] the EndPoint extensions are backwards compatible [11:03:00.0000] anyone know how to get all elements directly containing "foo" using XPath, and not any parent of a text node containing the string? [14:45:00.0000] /me considers posting hsivonen's NYT link to public-html [14:46:00.0000] hahaha [14:46:01.0000] /me notices that the 404 page on acid3.acidtests.org is coming out as the favicon in certain browsers [14:46:02.0000] unexpected bug! [14:47:00.0000] jgraham: what does the article say? [14:51:00.0000] Hixie: the article presents three cases where laws that are supposed to protect those who are worse off make their situation even worse my creating bad incentives for others [14:51:01.0000] a'h [14:51:02.0000] Hixie: that Americans with Disabilities Act gives an incentive to doctors not to take deaf patient [14:52:00.0000] Hixie: and that Mosaic law on money lending and agriculture are bad for the poorest in Israel [14:53:00.0000] Hixie: and how protecting habitats of endangered species creates an incentive for landowners to clear their land pre-emptively [14:53:01.0000] and how requiring 'alt' makes people add bogus values to shut up validators? :-) [14:55:00.0000] I guess the economists haven't examined HTML yet [14:56:00.0000] i think anyone who would realise the implications of that article are likely already convinced [15:09:00.0000] Hixie: I was considering saying that the critical take away message is that we have to operate a feedback loop: a) do things that we think are a good idea b) assess the _actual_ impact of those things c) make changes where it urns out things didn't have the desired impact [15:09:01.0000] We are at satge b [15:09:02.0000] s/satge/stage/ [15:09:03.0000] we're at all three stages on different parts of the spec [15:10:00.0000] but the above should go without saying for all spec development work [15:10:01.0000] and indeed, almost all work of any knid [15:11:00.0000] I guess that's not obvious to everyone. I really just want people to understand that stages b) and c) have to happen as part of HTML5 [15:11:01.0000] unless we want it to be no better than HTML 4 [15:11:02.0000] Hixie: evidence suggests that a lot of stuff that should go without saying given suitable economics or tecnology strategy background doesn't go without saying [15:12:00.0000] anyone any better way to achieve ? [15:12:01.0000] jgraham: i encourage you to advocate this; i'm just skeptical that there are people who are still unconvinced who can be convinced. [15:13:00.0000] gsnedders: Define better. You could do it non-recursively which would prevent stack overflow [15:13:01.0000] gsnedders: (sorry to bikeshed, but) wouldn't 'descendents' be a better name than 'children' in this case? [15:13:02.0000] i think a lot of people in the w3c think that standards _shouldn't_ have a feedback loop; that standards should be written by experts whose knowledge is by definition perfect and that therefore there cannot be anything to learn. [15:13:03.0000] jgraham: silly Python with it's stack :) [15:13:04.0000] Does anyone know if NYT print versions are registration-free? [15:14:00.0000] hober: ironically, I just renamed it from that. I'm still toying between the two. [15:14:01.0000] jgraham: more seriously, it copes with HTML 5 fine [15:14:02.0000] gsnedders: In practice you'd get a RecursionError or something [15:15:00.0000] (fwiw I'd write the code in the same way but I don't claim to be a particularly good programmer) [15:15:01.0000] /me thought of how to do it non-recursively, then promptly forgot [15:15:02.0000] and now all I think of has major flaws. [15:15:03.0000] Can people access http://www.nytimes.com/2008/01/20/magazine/20wwln-freak-t.html?_r=5&oref=login&ref=magazine&pagewanted=print ? [15:16:00.0000] jgraham: that one is behind login too [15:16:01.0000] however, having a referer might work [15:16:02.0000] i thought there was some magic "partner" argument one could set [15:17:00.0000] nope, tinyurling doesn't work [15:17:01.0000] Hixie, Acid3 test 69: "Statement on line 2502: Undefined variable: numberOfChars" [15:17:02.0000] Tantek had a magic tinyurl-like URL for that article that worked magically [15:18:00.0000] /me blinks [15:18:01.0000] /me found the article tantek's tweet [15:18:02.0000] The I18N WG expects every single JS author using selectors API to implement unicode case folding? [15:18:03.0000] Dashiva: or normalization [15:18:04.0000] or did I misread? [15:18:05.0000] case-insensitivity was a bad idea [15:18:06.0000] NYT link w/o login. http://www.nytimes.com/2008/01/20/magazine/20wwln-freak-t.html?_r=1&ex=1358485200&en=0d05099c03c97375&ei=5090&partner=rssuserland&emc=rss&pagewanted=all&oref=slogin [15:19:00.0000] /me now can't think of to do it non-recursively [15:19:01.0000] jgraham: http://www.nytimes.com/2008/01/20/magazine/20wwln-freak-t.html?_r=5&oref=login&ref=magazine&pagewanted=print works for me without login [15:19:02.0000] billmason: Thanks :) [15:20:00.0000] You're welcome. I use http://nytimes.blogspace.com/genlink to get those links, btw. Comes in handy. [15:20:01.0000] hsivonen: Yeah, that's the next point, I see [15:20:02.0000] anne: line 2502 is a comment [15:20:03.0000] annevk: fixed [15:21:00.0000] hmm [15:22:00.0000] ah yeah, now I get the error i was expecting [15:23:00.0000] not sure i'm going to debug it though, SVG is no fun [15:23:01.0000] hah [15:24:00.0000] https://bugzilla.mozilla.org/show_bug.cgi?id=379461 dbaron ftw! [15:24:01.0000] my memory says we fixed it too, so that makes two out of three browsers [15:25:00.0000] gsnedders: r=[]; c=[Node]; while c: c += c[0].childNodes; r.append(c.pop(0)) or something roughly like that, I think [15:27:00.0000] (Python isn't entirely good at one-liners, sadly) [15:28:00.0000] I think if you want to be really fancy you can do it without an explicit stack; basically for each node you go to the first child, if there is one, otherwise go to the next sibling otherwise go to the parent (but not back to he first child) [15:30:00.0000] That doesn't seem entirely necessary unless the current DOM is using more than 50% of your RAM and you can't afford to make another copy of it [15:30:01.0000] and then it's probably easier and cheaper just to stick another 2GB of RAM in so you won't have to worry about it [15:31:00.0000] :) [15:31:01.0000] Philip`: my recursive version is twice as fast going over HTML 5. peh. [15:31:02.0000] jgraham: yeah, that's what I was thinking of doing [15:32:00.0000] Does Python do something silly like not use double-linked lists for 'list'? [15:32:01.0000] double-linked lists? [15:32:02.0000] Philip`: list == array [15:33:00.0000] Hmm, that'd probably make .pop(0) quite expensive then [15:33:01.0000] gsnedders: You could investigate the deque class in the collections module [15:33:02.0000] Philip`: pop seems perfectly cheap [15:33:03.0000] Yeah, I guess pop(0) will be O(N) but pop(-1) will be cheap [15:34:00.0000] 97901 0.479 0.000 0.479 0.000 {method 'pop' of 'list' objects} [15:35:00.0000] (i.e., percall is 0.00 sec) [15:35:01.0000] gsnedders: Is that pop(0)? [15:35:02.0000] ya [15:36:00.0000] /me still suggests looking up deque [15:36:01.0000] http://pastebin.ca/905707 (top is non-recursive) [15:37:00.0000] from collections import deque; r=[]; c=deque([node]); while c: t=c.popleft(); c.extend(t.childNodes); r.append(t) maybe [15:39:00.0000] gsnedders: It looks like function call overhead is killing you [15:40:00.0000] deque is back up to speed [15:40:01.0000] specifically __getattr__ [15:40:02.0000] yeah [15:40:03.0000] that's what I thought [15:42:00.0000] /me still suggests using Psyco :-) [15:44:00.0000] Philip`: peh. at least I'm not actually parsing about ten/fifteen times like the real spec-gen [15:45:00.0000] (I think html5lib goes about twice as fast under Psyco) [15:45:01.0000] .extend() instead of += cuts out a lot of the function calls, so that makes the standard non-recursive one quicker [15:45:02.0000] /me still suggests lxml [15:45:03.0000] Philip`: (that would make a difference) [15:45:04.0000] /me tries to remember why he didn't use lxml [15:46:00.0000] cssutils does some fun "c = self.text.pop(0)" for every single character while tokenising stylesheets, which completely destroys performance because of the O(n) thing [15:46:01.0000] ...though it looks like that code is rewritten in the latest version, so maybe they fixed that problem [15:47:00.0000] /me really ought to get his head around big-O notation [15:51:00.0000] It's easy - O(1) means you're hiding the expensive operations on your data structure, O(n) means you're using the expensive operations on your data structure, O(n log n) means you're using trees, and O(e^n) means you're working on an interesting problem [15:51:01.0000] /me headdesks [15:52:00.0000] O(N^2) means the interviewer wants you to think of a different solution :) [15:56:00.0000] Can I get a serious explanation? [15:57:00.0000] http://www.google.co.uk/url?sa=t&ct=res&cd=1&url=http%3A%2F%2Fen.wikipedia.org%2Fwiki%2FBig_O_notation&ei=Qyi2R8u7IaWwwQHGwJDICg&usg=AFQjCNFsPl2CW_acj-E5rq3PZ17cwqaJ2g&sig2=K2R1u4iESPxXmFtCt8TmqQ [15:57:01.0000] er [15:57:02.0000] I didn't realise google did horrible redirects on its own urls thesedays [15:57:03.0000] takkaria: I've been trying to get my head around that article for a while. I've still failed. [15:57:04.0000] It rewrites the href just as you click on the link [15:58:00.0000] so it works alright if you right-click-and-copy-link before following the link [15:58:01.0000] N is the number of elements you feed in to the algorithm. O(foo(N)) means that as N changes the running time scales like foo(N) [15:58:02.0000] Oops, I broke it [15:58:03.0000] Oh, wrong window [15:59:00.0000] jgraham: and there's nothing more complex than that? [15:59:01.0000] gsnedders: working out what foo is for a given algorithm [15:59:02.0000] jgraham: well, obviously :) [15:59:03.0000] /me should probably leave this to someone with a CS background 2008-02-16 [16:00:00.0000] gsnedders: that's pretty much it, though there is math involved. Short conceptual summary, though.. Let n = 10^6. [16:00:01.0000] O(foo(N)) means the running time scales like foo(N) in the limit of large N [16:00:02.0000] (except it's stupid because for large N you're going to run out of memory) [16:01:00.0000] (unless you're Google) [16:01:01.0000] Philip`: how does it deal with memory too? [16:01:02.0000] (or me) [16:01:03.0000] :P [16:01:04.0000] /me needs more memory, badly. [16:01:05.0000] gsnedders: An O(1) algorithm will take some small number of operations... An O(n) will do ~1 million. O(n^2) one will do ~10^12, a trillion... An O(log n) one will do something to the order of 20, etc. [16:01:06.0000] Philip`: not really. Quadratic complexity shows up easily for very small input sizes. [16:02:00.0000] gsnedders: it can be applied to that, yes, though that's not as common. [16:02:01.0000] gsnedders: You can say you're measuring memory rather than processing time, like saying that an array of size n requires O(n) memory [16:02:02.0000] SadEagle: surely the number of ops. depends on the size of n? [16:02:03.0000] /me squints [16:03:00.0000] gsnedders: yes. but if you look at my examples, something that does about n^2 operations does a lot more work then something that does about n operations, when n is large enough. [16:03:01.0000] /me forgot you stated what value n was [16:04:00.0000] /me is too tired [16:04:01.0000] 1 million for illustrative purposes.. [16:04:02.0000] hmm, surely someone made a nice chart someplace.. Like probably on the WP article takkaria linked to. [16:04:03.0000] /me needs to go sleep [16:04:04.0000] g'nite [16:05:00.0000] goodnight [16:11:00.0000] Hixie, Acid3 loads xhtml.1 three times [16:11:01.0000] duh [16:11:02.0000] suck, i do [16:11:03.0000] fixed, it is [16:11:04.0000] thanks [16:14:00.0000] good, we still pass :) [16:15:00.0000] gsnedders: so, for O(f(n)), that implies that, given an input of size n, the algorithm runs in time/uses space in the range [c_1 * f(n), c_2 * f(n)], for all n > some k [16:15:01.0000] basically, long-run the runtime/space is proportional to f(n) [16:16:00.0000] oh, he left [16:16:01.0000] Strictly speaking that's theta, not O [16:16:02.0000] O is only an upper bound [16:16:03.0000] I am pretty sure he went to sleep :-) [16:16:04.0000] wrong; I said "in", not bounded by on both sides [16:17:00.0000] If it's in a range then that's bounding it on both sides [16:17:01.0000] (if it's a finite range, anyway) [16:17:02.0000] well, he did permit c_1 to be 0, but also c_1 wasn't constraint to be positive :-) [16:17:03.0000] har har [16:17:04.0000] /me pulls out clrs [16:17:05.0000] What's wrong with an algorithm that takes negative time? [16:18:00.0000] You're dragging down the theory with mere practical concerns [16:18:01.0000] There are algorithms based on answers from the future [16:18:02.0000] Does that count as negative time? [16:18:03.0000] f(n) = O(g(n)) iff \exists c > 0, n_0 >= 0. \forall n >= n_0. f(n) <= cg(n). [16:19:00.0000] What happens if your algorithm involves extreme gravitational forces and causes time itself to change as a function of n? [16:19:01.0000] Philip`: yeah, it's probably extraneous if you actually want O and not theta :-) [16:19:02.0000] er guys, i don't want to worry y'all, but it looks like your keyboards have got errors [16:19:03.0000] the text you're writing is all coming out with lots of weird punctuation [16:19:04.0000] latex ftw [16:20:00.0000] yeah, I did mistype the bounds for O [16:20:01.0000] I meant the right thing, tho! [16:20:02.0000] /me notes there have been far too few \ for latex [16:21:00.0000] c_0 is totally latex [16:21:01.0000] Intelligent sort is O(0), I wonder if you could make it negative somehow [16:21:02.0000] jwalden: Only in $ or some other math mode :) [16:21:03.0000] Hixie: would you prefer MathML? [16:22:00.0000] heh [16:22:01.0000] SadEagle: :-P [16:22:02.0000] (I guess that's one way of making me shut up :-) ) [16:23:00.0000] so, not to drag the topic of conversation back to html5 or anything, but i think i'm gonna have to make it so documents in session history always know what the last state object (from pushState) that was activated on that document is [16:32:00.0000] wow, i didn't mean to kill the conversation [16:42:00.0000] Hixie: What's the use case? [16:42:01.0000] making state objects work at all in the face of session history navigation [16:42:02.0000] the current algorithm is complete BS [16:43:00.0000] (this doesn't affect the APIs at all) [16:57:00.0000] jgraham__: very well expressed [17:09:00.0000] i checked in the aforementioned change [19:24:00.0000] hmmm [19:24:01.0000] /me investigates .query = '' and .hash = '' [19:34:00.0000] (s/query/search/. blame netscape people. probably brendan.) [19:55:00.0000] I wonder how the i18n WG realistically expects selectors api to require unicode case folding, given that an implementation of that in JavaScript is not trivial and would be ignored by most authros. [20:08:00.0000] what would it mean? [20:08:01.0000] case folding for what? [20:08:02.0000] prefixes? [20:08:03.0000] yes [20:08:04.0000] i thought the csswg had agreed to making that case-sensitive or something [20:08:05.0000] or ascii-only [20:08:06.0000] they somehow expect authors to implement that in their NSResolver [20:09:00.0000] the CSSWG agreed to make case sensitivity dependent upon the mechanism used to declare it [20:09:01.0000] oh [20:09:02.0000] so just make it sensitive [20:09:03.0000] and be done [20:09:04.0000] it basically is [20:09:05.0000] so what's the problem? [20:10:00.0000] did you read the mail from i18n WG on public-webapi? [20:10:01.0000] i read enough of it to know it wasn't my problem :-) [20:10:02.0000] /me looks again [20:11:00.0000] oh i see [20:11:01.0000] from the point of view of the UA, prefixes need to be treated case sensitively, so "foo" and "FOO" prefixes would need to be resolved separately. [20:11:02.0000] right [20:11:03.0000] but NSResolver can do whatever [20:12:00.0000] I would just remove all mention of the fact that the resolver can fold case [20:12:01.0000] since effectively that just means it's defining a set of prefixes [20:12:02.0000] yeah, so if the author implements a case insensitive NSResolver, they're effectively just declaring all variants of "foo", "Foo", "FOO", etc. [20:12:03.0000] right [20:13:00.0000] that removes your issue with the i18nguys [20:13:01.0000] I just need to figure out how to adjust the spec, without actually changing any requirements to keep them happy, and to respond to their mail [20:15:00.0000] hehe [20:20:00.0000] /me checks in his weird .search = '' issue [20:21:00.0000] I changed the first note to read: "Although prefixes are case sensitive, the implementation of the NSResolver object may internally handle them case insensitively. However, such an approach is effectively the same as declaring all case variants of the prefix with the same namespace URI." [20:22:00.0000] i think they'll still complain [20:22:01.0000] if they do, i'd just remove the note altogether :-) [20:22:02.0000] (after all, does it really help with anything?) [20:22:03.0000] brb [20:26:00.0000] I think I can resolve it in a way that they won't complain [20:27:00.0000] I think the original reason for the note was because Selectors required case insensitive prefixes [20:27:01.0000] but that's no longer the case [20:30:00.0000] right [20:40:00.0000] ok let [20:40:01.0000] us [20:40:02.0000] look [20:40:03.0000] at... [20:40:04.0000] the element [20:41:00.0000] anyone have any opinions i should consider in addition to the tons of feedback? [20:48:00.0000] keep it! [20:49:00.0000] ignore everyone's feedback that disagrees with my opinion [20:49:01.0000] yeah, reading this i'm convinced there are good use cases for it [20:49:02.0000] i think we should change the name though [20:49:03.0000] to what? [20:49:04.0000] is too short given that this won't be used that much [20:49:05.0000] i dunno [20:50:00.0000] maybe. would be way too misspelt... [20:51:00.0000] is ok [20:51:01.0000] highlight also suffers from the perception of being presentational, which tends to receive a lot of complaints [20:53:00.0000] yes [20:59:00.0000] http://lists.whatwg.org/pipermail/whatwg-whatwg.org/2007-February/009286.html is a great UI idea [21:05:00.0000] jesus, a lot of people mailed about [21:16:00.0000] do you mean the aural example of the UA saying "begin mark. (content) end mark.", or the UI of skipping between marked text? [21:16:01.0000] ui [21:18:00.0000] how do I deal with the unicode normalisation issue? [21:19:00.0000] I obviously can't require the NSResolver to normalise the prefix. Could I require the UA to normalise the prefix before passing it to the resolver? [21:21:00.0000] I would rather not, since that would create more surface area for bugs, and be really annoying for authors if one UA had a bug in its normalisation algorithm, resulting in a different string being passed from other UAs [21:35:00.0000] well [21:36:00.0000] it depends on whether you want to require that UAs normalise before parsing the selectors, or whether you want to require that they not normalise [21:36:01.0000] might be easiest to jus check what implemenations are doing [21:37:00.0000] yeah. [21:37:01.0000] weinig, what does webkit do? [21:38:00.0000] Lachy: sorry, i missed the beginning of the conversation. What is the question? [21:39:00.0000] Lachy: i wouldn't rely on implementors knowing what they do :-P [21:39:01.0000] Hixie: :( [21:39:02.0000] [06:25] I obviously can't require the NSResolver to normalise the prefix. Could I require the UA to normalise the prefix before passing it to the resolver? [21:39:03.0000] [06:27] I would rather not, since that would create more surface area for bugs, and be really annoying for authors if one UA had a bug in its normalisation algorithm, resulting in a different string being passed from other UAs [21:39:04.0000] weinig: when it comes to unicode normalisation, am i wrong? :-) [21:39:05.0000] Lachy: we have not implemented the NSResolver part of the spec yet [21:39:06.0000] Hixie: no, you are not wrong :) [21:39:07.0000] see :-D [21:40:00.0000] oops, missed [06:24] how do I deal with the unicode normalisation issue? [21:41:00.0000] weinig, see point 3 in http://lists.w3.org/Archives/Public/public-webapi/2008Feb/0095.html [21:42:00.0000] so, basically, the question is should the UA pass the prefix to the resolver in normalised form or exactly as it was used in the selector [21:43:00.0000] Lachy: xml requires normalisation of tag names right? [21:43:01.0000] Lachy: so the selector has to be normalised to that normalisation form anyway [21:43:02.0000] Lachy: so i'd say require that the selector be normalised to whatever normalisation form xml requires [21:44:00.0000] Lachy: hm, it seems that if we normalization of the prefix, that also means we have to normalize the selectors [21:44:01.0000] I'm not sure what XML says about normalisation [21:44:02.0000] check :-) [21:44:03.0000] Lachy: oh, I see Hixie just said that :) [21:44:04.0000] do you normalise the selectors? [21:44:05.0000] Lachy: I don't believe so [21:45:00.0000] Lachy: though, as Hixie noted, I may not know :) [21:45:01.0000] spec writing involves doing a lot of study work, examining implementations and other specs :-P [21:45:02.0000] /me nods [21:45:03.0000] /me is too lazy for that :-) [21:46:00.0000] could be a problem :-) [21:46:01.0000] I am going to say that we in fact do not normalize the string before parsing the selector [21:47:00.0000] you sure? i wouldn't bet in your js layer did it for you... [21:47:01.0000] er [21:47:02.0000] s/bet/be surprised/ [21:47:03.0000] /me checks there [21:48:00.0000] afk [21:48:01.0000] /me wonders how to write a test to see how browsers handle normalisation [21:49:00.0000] Lachy: that could obviate most of these questions :) [21:49:01.0000] the Selectors spec says nothing about normalisation [21:49:02.0000] I didn't think ECMAScript required normalization of strings [21:50:00.0000] I don't think it does either [21:51:00.0000] in that case, I really don't think we normalize, but I would still like to test it [21:51:01.0000] ap would know for sure [21:52:00.0000] can you find out from whoever knows and send feedback about it to public-webapi? [21:53:00.0000] Lachy: yes [21:53:01.0000] /me needs to discuss it with the devs at opera [21:53:02.0000] thanks [21:53:03.0000] no problem [00:27:00.0000] Lachy, no need to do normalization [00:27:01.0000] Lachy, nobody does that [00:33:00.0000] jgraham++ [00:37:00.0000] annevk: it's done more than you think [00:38:00.0000] (and there are multiple types of normalisation) [00:38:01.0000] (some of which make a lot of sense to do at certain points, others don't) [00:39:00.0000] however, it seems xml doesn't require normalisation [00:39:01.0000] so it would seem best to not do it here [00:39:02.0000] why am i doing this research :-P [00:40:00.0000] /me goes back to doing the research for his own spec :-P [00:41:00.0000] I'd be interested to know where NFC is applied in specifications authors deal with and authors provide the strings [00:42:00.0000] NFKC is applied when creating the punycode form of IDN, iirc [00:44:00.0000] ah, yeah [00:45:00.0000] I guess that's because i18n experts have been actively involved in that [00:45:01.0000] it would be a security disaster if it wasn't [00:45:02.0000] it's bad enough as it is [00:46:00.0000] true [03:14:00.0000] heycam, any chance you can make an update for test 69 in Acid3? the summary for instance is bogus now [03:14:01.0000] well, partially bogus [03:28:00.0000] Lachy: Hi there, with http://lachy.id.au/log/2005/05/script-comments in the "The Correct Method for XHTML" what's the // in //]]> for? contrasting with http://www.w3.org/TR/xhtml1/#h-4.8 [03:29:00.0000] that applies when serving as text/html, as the article should point out [03:29:01.0000] it's to comment it out so it doesn't cause a JS error [03:30:00.0000] oh, oops [03:30:01.0000] yeah, it's not needed. [03:30:02.0000] but harmless [03:30:03.0000] What's the browser support status of http://contentlabel.org/ ? [03:31:00.0000] (the technology advocated--not the site itself) [03:34:00.0000] Since it uses RDF and believes it can some how use it to increase trust (highly unlikely), I'm going to guess zero support [03:37:00.0000] I http://contentlabel.org/firefox-extension-that-reads-semantic-labelled-sites/ [03:38:00.0000] so it's vapourware except for their own Fx extension [03:39:00.0000] (which isn't as bad as some things, I guess) [03:40:00.0000] it's not clear to me what the motivation is. 1) Protecting children? 2) Causing an illusion of porn self-regulation in order to lessen government interest in statutory regulation 3) find a nail for a SemWeb hammer. 4) something else? [03:41:00.0000] It's possible to filter the web. kthxbai. [03:41:01.0000] hsivonen: Sell Segala trust labels I suspect. [03:42:00.0000] webben: hmm. yeah, that's a very plausible explanation [03:43:00.0000] I think currently their main customer is O2. [03:44:00.0000] how useful segala certification is perhaps open to doubt however: http://www.blether.com/archives/2006/06/da_vinci_code_t.php [03:44:01.0000] "Using W3C standards such as RDF makes mass adoption for Content Labels more seamless in our opinion." hmm [03:45:00.0000] well, using standards /should/ help adoption (at least, assuming the standards are workable) [03:47:00.0000] and there are plenty of tools that can extract stuff from RDF [03:47:01.0000] All W3C standards are totally widespread. [03:47:02.0000] gsnedders: well, all w3c standards are at least open to implementors [03:47:03.0000] webben: That's not overly relevant if nobody uses them, though [03:48:00.0000] though not all are implementable [03:48:01.0000] like I said: "assuming the standards are workable" [03:48:02.0000] RDF has been implemented, so it is workable. [03:50:00.0000] RDF is too complicated and only solves imaginary problems [03:50:01.0000] indeed, it's even implemented in browsers to some degree. [03:54:00.0000] 'saying stuff about stuff in a machine-readable manner" is an imaginary problem? [03:55:00.0000] or do you mean the added complications only solve imaginary problems in that task? [04:01:00.0000] hmm. looks like Segala is into the MobileOK stuff as well [04:01:01.0000] I don't like MobileOK at all [04:02:00.0000] how come? [04:02:01.0000] /me hasn't looked at it. [04:03:00.0000] http://www.w3.org/TR/mobileOK-basic10-tests/ ... interesting worked on by a google fellow and someone who used to work at Segala [04:03:01.0000] webben: the Mobile OK thing is based on the assumption that mobile browsers suck and that content needs to be dumbed down [04:04:00.0000] well, that's probably true of a lot of current deployment [04:05:00.0000] hopefully not so true in future [04:05:01.0000] webben: The mobile browsers I use can do much more than what is assumed by Mobile OK [04:05:02.0000] yes, I'd tend to assume web-loving folk would use better clients than average [04:05:03.0000] webben: the mobile browsers I use are based on Gecko, Opera and WebKit [04:06:00.0000] yes, but those are very much a new breed aren't they? [04:07:00.0000] guess it depends how fast such devices get rolled out into growth markets [04:07:01.0000] webben: I have about zero sympathy for bad products when at least three better ones are available [04:08:00.0000] webben: Opera Mini works on remarkably sucky phones [04:08:01.0000] opera mini is good [04:08:02.0000] dunno what the state of its localization is though [04:11:00.0000] http://operawatch.com/news/2007/11/opera-mini-translated-into-over-50-languages-15-more-to-come-later-this-month.html [05:03:00.0000] http://lists.w3.org/Archives/Public/public-html-comments/2008Feb/0011.html [05:03:01.0000] I'm not sure how I should react [05:05:00.0000] http://annevankesteren.nl/2007/04/html-red-pill [05:23:00.0000] /me wonders why Opera uses 1005MB of memory and then takes two minutes to exit [05:28:00.0000] well, replied anyway. [05:42:00.0000] "The reference image for Acid3 as rendered by WebKit r30069." wow, Wikipedia, thanks for telling me that this *reference image* screenshot was done with WebKit, and with a revision number too! [05:48:00.0000] you can fix that :p [05:49:00.0000] I wonder why they took out the screen shots of beta releases though [05:53:00.0000] Philip`: I have that constantly [05:56:00.0000] Hope it's fixed with 9.5 [05:59:00.0000] for real? [05:59:01.0000] /me doesn't have that problem at all [05:59:02.0000] I haven't installed 9.2 [06:00:00.0000] Perhaps that's the problem [06:00:01.0000] Steps to reproduce: Visit lots of web pages for a couple of days [06:00:02.0000] I had a laptop crash, bought a new one, and copied the Opera dir from my other disk [06:00:03.0000] (I'm using 9.2something) [06:00:04.0000] (which is one of the best things Opera provides) [06:00:05.0000] (It's not a problem with e.g. Flash, because I've got plugins disabled since Flash didn't work anyway) [06:01:00.0000] (and we both like parentheses) [06:02:00.0000] (They are an aesthetically pleasing curve) [06:02:01.0000] Perhaps it's Vista [06:02:02.0000] /me is using Linux [06:03:00.0000] I don't think it's Vista [06:03:01.0000] :p [06:03:02.0000] Mem usage here is around 200MB, even after closing Opera [06:03:03.0000] And it drops after a few minutes [06:04:00.0000] And if you restart it when it's not closed down yet, I get a mail init error [06:04:01.0000] s/you/I [06:05:00.0000] Anyway, I'm still alive, so it's probably not a big issue :) [06:13:00.0000] Philip`, I almost never reboot Opera and I don't encounter this problem (I'm on Linux too, Opera 9.2x) [06:13:01.0000] hopefully it's better in 9.5 [06:18:00.0000] annevk: you also use M2? [06:18:01.0000] yeah [06:19:00.0000] and IRC [06:19:01.0000] Hmm [06:19:02.0000] 9.5 also takes a while to close down for me, on a different machine [06:45:00.0000] hmm, I can project my laptop on my TV, but not both at the same time :( [06:55:00.0000] The laptop display switches off when you connect the TV? [06:57:00.0000] some distorted display, now it sort of works [06:58:00.0000] it's not perfect though as my TV only shows 2/3 of the screen [06:59:00.0000] tv has a res of 1360x768 and the laptop has 1680x1080 [07:00:00.0000] but I can watch movies now, sort of [07:01:00.0000] (oh also sound isn't patched through, but I guess that's to be expected with a VGA connector :) ) [07:06:00.0000] annevk: hmm. 1080p laptop. nice. what are the physical dimensions? [07:07:00.0000] 1080p is usually 1920x1080, so it's a bit misleading to call it that [07:08:00.0000] oh. right [07:11:00.0000] /me discovers that Ubuntu's do-release-upgrade is not foolproof, as it failed with "Could not install the upgrades. The upgrade aborts now. Your system could be in an unusable state." [07:11:01.0000] I went to a home eletronics store the other day. they had a wall full of flat 720p and 1080p TVs and the same blu-ray 1080p feed into all of them [07:11:02.0000] I couldn't tell the difference between 1080p and 720p [07:12:00.0000] and I couldn't tell the difference between 890 euro models and 1700 euro models [07:12:01.0000] makes one go hmm. [07:15:00.0000] The only TV I watch is over streaming Flash and costs me £0 (plus no TV Licence fee) and the quality seems perfectly adequate :-) [07:16:00.0000] But movies are different, since quality is often much more important there [07:20:00.0000] Philip`: exactly what drugs were you on when you decided you wanted to do Element.setAttribute(":", ...) anyway? :-P [07:21:00.0000] jwalden: I just wanted to do things http://philip.html5.org/tools/parser/?%3Ctest%3Atest%20test%3Atest%20%3Atest%3E and detect when I was about to create an invalid attribute so I could get rid of the offending characters [07:22:00.0000] s/things/things like/ [07:23:00.0000] but that turns out to be impossible because "invalid attribute" varies between browsers, rather than following any standard [07:26:00.0000] not like anyone actually specified that at the time browsers implemented it, really [07:26:01.0000] /me wonders why hand-waving is so prevalent in specs [07:26:02.0000] I'd be happiest if nothing was considered an invalid attribute, since otherwise it's impossible to write a conforming HTML5 parser using the JS DOM [07:27:00.0000] Element.setAttribute(" foo='bar' baz", "quux") [07:27:01.0000] attribute injection! [07:29:00.0000] Wouldn't it be easier to do Element.setAttribute('foo', 'bar') if you're able to control the arguments? :-p [07:29:01.0000] not necessarily, if the first argument was checked against a blacklist [07:30:00.0000] I don't presume to expect rational behavior in random web apps [07:30:01.0000] cf. ":" ;-) [07:31:00.0000] IE and Opera users would already be vulnerable if a web app did that [07:31:01.0000] hence web apps can't do that [07:32:00.0000] hence it's not a security problem :-) [07:32:01.0000] /me defines security problems out of existence [12:38:00.0000] hello [12:38:01.0000] Fiboknight, hi [12:38:02.0000] how may i get access to your irc log script? [12:38:03.0000] ask krijnh [12:38:04.0000] he runs it [12:39:00.0000] oh ok [12:39:01.0000] are you a dev as well? [12:40:00.0000] I was a web developer [12:42:00.0000] cool [13:53:00.0000] do adjacent text nodes have implicit whitespace between the two? [13:54:00.0000] Lachy: see the email I sent? [14:08:00.0000] they do not [14:11:00.0000] that's what I expected [14:11:01.0000] Why would (in Python) `else Element.textContent:` cause a syntaxerror? [14:11:02.0000] why not? [14:12:00.0000] gsnedders: elif [14:12:01.0000] hasather: oh duh. [14:12:02.0000] /me headdesks [14:12:03.0000] thx [14:12:04.0000] This is not the syntax you're looking for *handwave* [14:13:00.0000] Dashiva: hey! I'm allowed to be dumb at times! :) [14:13:01.0000] Dashiva: don't be mean to the kid! [14:14:00.0000] I'm just trying to lighten the mood with some humor [14:15:00.0000] And I'm being totally serious :) [14:15:01.0000] (obviously) [14:15:02.0000] [14:15:03.0000] Parse error [14:15:04.0000] Dashiva: what? you think you need a start tag! Silly XML making you think such things. [14:15:05.0000] is an optional start tag [14:16:00.0000] like , , and [14:16:01.0000] /me actually saw a real site which omitted the opening head tag, but not the closing one, a few weeks ago [14:16:02.0000] gsnedders: Since when is IRC sent as text/html? :P [14:17:00.0000] Dashiva: it's text/plain;charset=unknown, so why any parser error ever? :P [14:18:00.0000] I have to parse your writing to understand it [14:18:01.0000] peh! [14:18:02.0000] parsing is overrated. [14:19:00.0000] wow, the whatwg copy of html5 sure has dropped down the rankings [14:19:01.0000] Hixie: talking of HTML 5, you have any need for ul.toc on _every_ nested ul in the TOC? [14:20:00.0000] actually, why isn't it an ol? [14:21:00.0000] no idea on either point [14:21:01.0000] it should be an ol, no? [14:21:02.0000] probably [14:22:00.0000] /me goes back to designing an algorithm to actually create the TOC [14:22:01.0000] the html5 spec has one of those :-) [14:23:00.0000] though it's hard to understand [14:23:01.0000] it does? [14:23:02.0000] oh, that one [14:23:03.0000] /me totally forgot [14:25:00.0000] the creating an outline section? [14:26:00.0000] gsnedders, the proxy script you sent me doesn't correctly pass the username and password [14:26:01.0000] Lachy: I just realised that [14:28:00.0000] /me has an imp. of the HTML 5 algorithm somewhere but doesn't recommend it [14:31:00.0000] /me emails Lachy the "don't write PHP scripts quickly thinking you can write them fine" edition. [14:31:01.0000] jgraham: the alg, or the results? [14:32:00.0000] Hixie: What do you mean results? [14:32:01.0000] I think the algorithm does the right thing [14:32:02.0000] (more or less) [14:32:03.0000] that's what i meant by results [14:34:00.0000] I mostly meant that I don't recommend my implementation, although I have no particular love for the way the algorithm is expressed either [14:34:01.0000] /me is scratching his head at parts of the description [14:35:00.0000] jgraham: k [14:35:01.0000] bbl [14:35:02.0000] bye [14:37:00.0000] (I actually meant creating the actual TOC output, not creating it internally, but hey) [14:39:00.0000] gsnedders, you still didn't get it quite right :-) [14:39:01.0000] but I fixed it [14:39:02.0000] Lachy: what now? [14:39:03.0000] oh. ). [14:39:04.0000] yeah [14:39:05.0000] it all works [14:40:00.0000] A change so minor that "nothing" can go wrong. But of course it now causes parse errors. Typical :) [14:46:00.0000] /me waves g'nite [15:58:00.0000] please someone help me with an irc logger 2008-02-17 [16:03:00.0000] /me sees that we are second on http://www.google.com/search?q=irc+log [16:03:01.0000] or fourth, if I use a different browser [16:04:00.0000] Hooray for personalised search results [16:04:01.0000] I'm not sure this is the best place to find out about IRC logging, though [16:08:00.0000] thanks [12:15:00.0000] Philip`: Yeah, I should put Google Ads on the logs ;) [13:40:00.0000] http://lists.w3.org/Archives/Public/public-html-comments/2008Feb/0014.html [13:40:01.0000] what's the actionable feedback? [13:44:00.0000] jgraham: what lens you use for those two photos you just put on Flickr? [13:48:00.0000] hsivonen: Hmm. Well, Frank appears to want to be able to force a mime type in the validator. He also seems to want validator to test URIs are URIs not IRIs. [13:50:00.0000] would Validator.nu users be better off in general if I made the XHTML 1.0 schemas really require URIs instead of IRIs? [13:51:00.0000] if XHTML 1.0 requires URIs not IRIs and validator users are trying to validate XHTML 1.0 then yes. [13:53:00.0000] I guess it depends on whether you consider external refs to be independently updateable [13:54:00.0000] like the it wouldn't do anyone any good if Validator.nu used Unicode 2 insteado of Unicode 5 [13:54:01.0000] although W3C specs in general haven't yet been updated to point to Unicode 5.0 [13:57:00.0000] you mean the IRI requirement is set in a different spec? [13:58:00.0000] I guess that would depend on the spec. [13:58:01.0000] If it's debatable, then you'd need to give users an option. [13:58:02.0000] or seek clarification from whoever's responsible for the spec in question [13:58:03.0000] webben: I mean if a spec points to URIs when IRIs weren't specified, should software developers treat IRI as an update of URI as far as markup language attribute values go [13:59:00.0000] hsivonen, I don't think it's worth bothering about details that should've been errated in specs long ago [13:59:01.0000] (and where the spec in fact hints that the situation is going to change) [13:59:02.0000] gsnedders: Cannon 60mm EF-S 2.8 Macro [14:00:00.0000] hsivonen: only the group responsible for maintaining the XHTML 1.0 spec people are trying to validate to can answer that one way or the other. [14:00:01.0000] I guess that's Pemberton's lot atm. [14:00:02.0000] jgraham: I was thinking of getting that sometime (in a while, as I want a telephoto lens first) [14:01:00.0000] hsivonen: I guess the practical answer depends on whether any software depends on URIs-only [14:02:00.0000] jgraham: really old browsers do [14:02:01.0000] gsnedders: You should consider the 100mm Macro too; I didn't get it because this one was cheaper and someone else was paying [14:02:02.0000] jgraham: I really don't do much macro photography, so I really wouldn't want to get a very expensive one [14:03:00.0000] hsivonen: s/software/current software/ then. [14:03:01.0000] I'm inclined to fix this in documentation and claim that the schema isn't XHTML 1.0 but XHTML 1.0 plus IRI [14:04:00.0000] gsnedders: Well with lenses "expensive" means very different things to different people :) [14:04:01.0000] jgraham: that's true :) [14:12:00.0000] jgraham: for a start: L lenses are madly expensive :) [14:37:00.0000] i have way too much fun making my examples in the spec refer to things i like [14:41:00.0000] Hixie, what examples are you adding to the spec now, and what do they refer to? [14:41:01.0000] (now called ) [14:41:02.0000] http://omniplex.blogspot.com/2007/05/fictitious-u1e9e-character-endorsed-by.html [14:41:03.0000] and you'll have to see if you can guess them :-) [14:41:04.0000] ok [14:42:00.0000] Unicode needs a mechanism that stops the implementation cost of a new character from being an externality from the proposer point of view [14:43:00.0000] I still never figured out that ΑΒΓ company alt text example [14:44:00.0000] though, I found out the XYZ company example is possibly a reference to one of your old test cases on your site [14:47:00.0000] that's all it was iirc [14:49:00.0000] the ΑΒΓ company one too? I haven't seen that test case yet. Though, it'll probably show up in spartan one day [14:49:01.0000] i think it's just a different company with the same kind of name [14:49:02.0000] but i could be mistaken [15:01:00.0000] /me checks in the new element [15:01:01.0000] is that m with a longer name? [15:02:00.0000] yes [15:02:01.0000] the list of people bcc'ed on this e-mail is basically the list of people who contribute to the spec on a regular basis [15:02:02.0000] it's quite a long list [15:02:03.0000] like, 20+ names [15:03:00.0000] woot, that cleared out 74 e-mails in one go [15:03:01.0000] woah, that's one long email! [15:04:00.0000] anything worth reading in it? [15:04:01.0000] it's basically summarised at the top [15:04:02.0000] i didn't add much commentary [15:05:00.0000] /me optimistically deletes the "input-for-whatwg-semantics-phrasing-m" folder [15:17:00.0000] so the next topic is [15:17:01.0000] should it be only for citations, or for any title of work? [15:17:02.0000] you could add an example that uses the irc logs which are hilited in exactly the suggested way [15:17:03.0000] Hixie, only titles of work? [15:18:00.0000] is that a vote or a question? :-) [15:19:00.0000] people use it for titles of work in practice. I see no reason to restrict that [15:19:01.0000] Hixie, it's what hsivonen and DanC want I think [15:19:02.0000] (and yeah, irc might be a good thing to add) [15:20:00.0000] brb, intermission shift [15:20:01.0000] I've used it for citations, but only because I wanted to comply to HTML4... [15:20:02.0000] /me is shocked [15:21:00.0000] /me giggles at yet another Stargate reference in the spec :-) [15:22:00.0000] Will the conspiracy of light manage to sneak itself in there? [15:22:01.0000] Hixie, s/just under 39 minutes/just over 38 minutes/ [15:24:00.0000] the only ones to last longer than a few seconds past 38 minutes had powerful energy sources keeping them open [15:26:00.0000] Hixie, s/and a large gravity well/or a large gravity well/ [15:36:00.0000] /me reads Hixie's email [15:39:00.0000] "The semantic in question isn't the kind of thing I would imagine would fit the microformat ethos." -- which is correct [15:40:00.0000] which semantic? [15:41:00.0000] (which was ) [15:43:00.0000] tantek http://www.w3.org/html/wg/html5/#the-mark [15:44:00.0000] hmm.. last time i tried viewing the html5 spec it locked up firefox. not sure i want to click that link ;) [15:51:00.0000] /me wonders what the use case of marking titles is [15:54:00.0000] tantek, here's the multipage version http://www.whatwg.org/specs/web-apps/current-work/multipage/section-phrase.html#the-mark [15:56:00.0000] but even the single page version shouldn't lock up firefox. I look at it all the time without any problems. It generally loads in about 5-10 seconds for me [15:58:00.0000] At one point I was having an issue with the spec and a ff extension, but I don't recall which one 2008-02-18 [16:01:00.0000] Lachy: a few seconds above 38 is a few seconds under 39 :-P [16:01:01.0000] and i think the conditions include A and B is correct [16:02:00.0000] To clarify, by "marking titles" above, I meant marking them up using rather than anything to do with [16:03:00.0000] jgraham: "my favourite book series is The Night Dawn's Trilogy." [16:04:00.0000] the HTML specs have been consistent in making cite mean citation not just title e.g. "The CITE element is used to indicate the title of a book or other citation" (HTML 2.0). "CITE used for citations or references to other sources" (HTML 3.2). [16:04:01.0000] Hixie: So, apart from the italics for visual presentation, why should I care that it's a book title rather than something else? [16:04:02.0000] it's also worth noting that not all titles are italicized [16:05:00.0000] jgraham: "So, apart from the italics for visual presentation, why should I care that it's emphasis rather than importance?" [16:05:01.0000] and if one were to restrict cite to titles that happen to be italicized in a particular citation style, it's hard to see any advantage of cite over i [16:05:02.0000] (Do non-visual UAs do something different for compared to e.g. )? [16:05:03.0000] I'm not sure what they do. [16:06:00.0000] csarven, agreed, the semantic of is better suited to HTML than a microformat. it sounds like a special purpose element IMHO. [16:07:00.0000] Probably the most interesting consuming agent to look at would be Zotero, see if they do anything with it. [16:07:01.0000] (or would like to do anything with it) [16:07:02.0000] Hixie: in the case of something generic like emphasis v importance, I can imagine that there will be a presentational distinction in all media [16:07:03.0000] and many uses [16:08:00.0000] jgraham, if a UA knows semantically that something is the name of a book or other such reference, it may be possible for it to look it up in various citation references, or points of viewing/previewing/purchase. [16:08:01.0000] jgraham: i guess the real question is whether we should define "citation" as only something to which more than a passing reference is made, or whether any reference to another work is enough to be considered a citation [16:08:02.0000] tantek: Is that realistic, given how non-unique titles of works are? [16:08:03.0000] Hixie: Agreed that a big problem with CITE is clarifying what citation is. e.g. not quotation [16:09:00.0000] tantek I agree. It appears to me more of an element that is used for visual or interaction then a semantic representation of some content. I can understand the real-world use cases but that tends to get into a gray area IMO [16:09:01.0000] Hixie, the series always said 38 minutes, and most close closer to 38 minutes than to 39 minutes. (There was only one that lasted 38:34 in Chain Reaction, but that extra time was because the planet became a ball of plasma) [16:09:02.0000] jgraham, seems to work well enough in practice, humans communicate names of movies, songs, books all the time and are able to get value out of such communication. [16:09:03.0000] jgraham: i would imagine it is realistic in the context of a specific site with some specific JS to do e.g. referrals to amazon [16:09:04.0000] Note that in academic circles, you cite people as sources not just works. [16:10:00.0000] Lachy: agreed [16:10:01.0000] Hixie, it needs to say "or" instead of "and" because the it only requires one or the other, not both [16:10:02.0000] tantek: It's not clear to me that all things that work well in two-way human-human conversation translate to good UI for browsers [16:10:03.0000] Lachy: (38 is a number in stargate used in much the same way as 47 in startrek) [16:11:00.0000] Lachy: the list includes A and B, i don't say "it can happen if A and B" [16:11:01.0000] jgraham, we don't need "all things", just 80/20. no need to boil the ocean. [16:11:02.0000] /me covers his ears and starts yelling "NOT LISTENING!" :-) [16:11:03.0000] To me a citation is specifically a work that is drawn on by the current work and should be referenced for background [16:12:00.0000] Hixie, is that also true for wormholes between supergates? [16:12:01.0000] /me chooses to believe the 38 minute limit can be explained by quantum mechanics [16:12:02.0000] :-) [16:12:03.0000] tantek: Still, I don't recally any of the systems for "helpfully" adding referral links to webpages based on keywords ever hitting anything like an 80/20 point for me [16:12:04.0000] bbl [16:14:00.0000] jgraham, true, invisible meta keywords FAIL. visible tags FTW. also human-based folksonomy currently wins over "helpful" AI autotagging. [16:15:00.0000] /me -> sleep [16:15:01.0000] NLP is trying to solve a much greater problem which probably won't really be solved anytime 'soon' [16:17:00.0000] (At least in digital computing) [16:17:01.0000] csarven, true. just let the singularity solve the NLP problem. [16:18:00.0000] (sorry, had a shift) [16:19:00.0000] tantek: no iea [16:19:01.0000] Lachy: long story short, it's just an example. :-P [16:20:00.0000] anyway. we clearly must keep , and since we're keeping it, how to define it is key [16:20:01.0000] systems referring users to for-pay systems are probably a bit less useful (and of course, a bit more profitable) than something referring users to e.g. a university library resource resolver [16:20:02.0000] i'm not going off for a few hours. got another shfit and then i'm off to the other side of the stage and then to dinner. [16:20:03.0000] bbl. [16:20:04.0000] i'll read any insights you have upon my return :-) [16:21:00.0000] "I also have some kittens" - that makes me wonder whether "kittens" would be acceptable (from a cleverer search engine that detects words with the same basic meaning, rather than doing substring matches, and it might get young cat too) [16:23:00.0000] (I believe it is perfectly acceptable, but the example makes me wonder anyway, so maybe the example could say kittens to be clear that it's not meant to be strict about anything) [16:23:01.0000] it's intriguing folks aren't sure CITE is 100% semantically appropriate for citations in zotero's output: http://forums.zotero.org/discussion/2169/ [16:24:00.0000] http://en.wikipedia.org/wiki/Template_talk:Citation also interesting [16:25:00.0000] "
int x = 1
Error: expected semicolon; got EOF" - I guess that wouldn't work, which seems a problem if you want to use this for highlighting sections of code [16:28:00.0000] Is it safe to copy xkcd text without being affected by Creative Commons? [01:59:00.0000] jwalden, the security section already deals with cross-origin stuff [01:59:01.0000] jwalden, also for setting fillStyle and strokeStyle [02:01:00.0000] sure; it doesn't, however, say that drawImage with a different-origin image should not throw [02:02:00.0000] Is it not possible to access the .complete of a different-origin image? [02:03:00.0000] jwalden, it does [02:04:00.0000] where? [02:04:01.0000] http://www.whatwg.org/specs/web-apps/current-work/multipage/section-the-canvas.html#security1 [02:05:00.0000] that just says origin-clean is set to false, not that it doesn't throw [02:05:01.0000] oh right, sorry [02:05:02.0000] why should drawing fail? [02:05:03.0000] there's no risk there [02:07:00.0000] it can draw or not draw depending on whether the resource is a valid image, it just seems reasonable to say you shouldn't be able to differentiate the two for a different-origin image [02:07:01.0000] .complete doesn't say anything about origins [02:07:02.0000] which might make the concern moot, now that you mention it [02:07:03.0000] onerror fails for non same-origin images too [02:07:04.0000] euh, fires [02:07:05.0000] s/fails/fires/ [02:09:00.0000] hrm [02:09:01.0000] so basically, worrying about this is probably not worthwhile, I guess [02:10:00.0000] yeah [02:11:00.0000] certainly wouldn't be the place to fix the issue [06:25:00.0000] for 'setAttribute(a, v1); v2 = getAttribute(a)' v1 != v2 when a is style in Gecko (for some values of v1) [06:26:00.0000] oops [06:36:00.0000] anyone going to www2008.org in Beijing? i need some motivation [06:37:00.0000] my manager prolly [06:38:00.0000] aka chaals [06:38:01.0000] annevk, do you know if comments like this one, which don't require any action, need to be recorded in the disposition of comments? http://lists.w3.org/Archives/Public/public-webapi/2008Jan/0021.html [06:39:00.0000] explicit endorsement is never bad, i'd add it [06:41:00.0000] done [06:48:00.0000] I wonder how I can deal with this comment http://lists.w3.org/Archives/Public/public-webapi/2008Feb/0104.html [06:49:00.0000] I don't know what the spec should say about the issue [06:49:01.0000] of NSResolver mutating the DOM [06:51:00.0000] I wonder if I could say that, due to the possibility of the NSResolver mutating the DOM, that all prefixes must be resolved before looking for matching elements [06:52:00.0000] that way, if NSResolver did modify the dom, then all modifications would have occurred before the UA began looking [06:55:00.0000] " [06:55:01.0000] Now maybe you're actually requiring that the number of calls to the NSResolver [06:55:02.0000] for any given selector and initial DOM tree is bounded in the face of all [06:55:03.0000] possible mutations by the NSResolver and that hence the DOM will at some point [06:55:04.0000] stabilize and it will be possible to return the things the spec requires be [06:55:05.0000] returned. But if that's a constraint you want to place on implementations, you [06:55:06.0000] should probably spell it out clearly." [06:55:07.0000] is what I think the spec already says [06:55:08.0000] but we could add a paragraph that spells it out clearly [06:55:09.0000] annevk: met charges at MWC2008 in Barcelona, aka HELL [06:55:10.0000] Lachy, indeed [07:03:00.0000] Lachy: s/exersise/exercise/ in the note about case-mapping [07:04:00.0000] fixed [07:21:00.0000] annevk, how does this sound: [07:21:01.0000] Elements returned by these methods must include only matching Element nodes, which were present in the document after all namespaces that the implemented needed to resolve, have been resolved. [07:21:02.0000] Note: This is to ensure that any DOM modifications performed by a misbehaving namespace resolver have occurred prior to matching any elements. [07:21:03.0000] s/implemented/implementation/ [07:23:00.0000] The first sentence doesn't sound right - if someone queried for "a b" then the resolver moved the 'b' outside the 'a' (but kept it present in the document elsewhere), it's not clear that it shouldn't be matched [07:24:00.0000] s/charges/chaals/ woops [07:26:00.0000] Philip`, then it would no longer be a matching element node [07:28:00.0000] I'm not sure how I can make it any clearer. Elements are only determined to match at the they are evaluated with the selector [07:28:01.0000] It was a matching element node (before the resolving), so it is one of the matching nodes which is still present in the document [07:28:02.0000] but the implementation wouldn't know that it was matching until it started looking for it [07:28:03.0000] *still present in the document after the resolving [07:32:00.0000] hmm. I just realised my current defintions of element.querySelector require the matching elements to be within the document, but that wouldn't be the case if the element itself wasn't in the document [07:33:00.0000] oh, no they don't [07:33:01.0000] I misread it [07:34:00.0000] "'Matching Element nodes' are the Element nodes that match after all NSResolver calls have been made" or something like that? [07:35:00.0000] Incidentally, "The implementation must process the selectors according to the grammar of Selectors ([Selectors], section 10)." should have small-caps "must" [07:35:01.0000] (as should "the implementation must act as if the nsresolver argument was set to null") [07:46:00.0000] fixed [07:47:00.0000] Lachy, I'm away for a while, just check in what you think is appropriate, i'll try to scream whenever it's not :p [07:50:00.0000] Elements returned by these methods must include only Element nodes that are present within the document or the element’s subtree that match the group of selectors after all namespaces that need to be resolved by the implementation, have been resolved. [07:51:00.0000] Philip`, how's that now? [07:51:01.0000] s/subtree that/subtree and/ [07:52:00.0000] this is the corrected version: [07:52:01.0000] Elements returned by these methods must only include Element nodes that are present within the document or the element’s subtree, and match the group of selectors after all namespaces that need to be resolved by the implementation, have been resolved. [07:55:00.0000] Lachy: I think your commas are in the wrong places [07:56:00.0000] Actually I just think that the last comma is wrong [07:56:01.0000] (i.e. you should delete it) [07:57:00.0000] ok [08:04:00.0000] hmm. I need to think about this some more. I'm not sure if it completely addresses Boris' comment [08:07:00.0000] Lachy: Sounds reasonable to me now [08:19:00.0000] should the presence of a ::selection rule set supress the default styles for selections? [08:20:00.0000] s/rule/ruleset/ [08:21:00.0000] zcorpan: I think it's browser-specific how they combine [08:25:00.0000] /me thinks it should be defined [08:34:00.0000] zcorpan: the expected model would be different if the default selection style is something that can be itself achieved through style rule vs. if it is a totally custom rendering of some kind [08:35:00.0000] ::selected { display: none } /* Anti-Copy-And-Paste Script, copyright 2008 www.coolscripts.com */ [08:36:00.0000] currently both opera and firefox supress default styles in the presence of ::selection (or ::-moz-selection) [08:36:01.0000] Philip`: display doesn't apply [08:36:02.0000] Oh [08:36:03.0000] Philip`: or it's not required to apply [08:36:04.0000] the selection pseudo-element allows a limited set of style properties [08:37:00.0000] in Safari, I think the background drawing still happens unless you explicitly suppress it [08:37:01.0000] ::selected { color: expression(alert("Anti-Copy-And-Paste!")) } [08:38:00.0000] supressing the default styles is different from how css usually works, so not what i had expected [08:40:00.0000] oh wait, opera doesn't supress unless there's a color or background property, it seems [08:43:00.0000] Safari models it as a magical background-color property basically, although there isn't an explicit rule for it in html4.css [08:44:00.0000] othermaciej: does Apple disclose the mechanism of Safari content blocking when parental controls are enabled? [08:45:00.0000] hsivonen: I'm not sure [08:45:01.0000] othermaciej: ok [08:45:02.0000] hsivonen: I believe the parent can set up a whitelist via bookmarks, but I don't know if there is more to it [08:47:00.0000] othermaciej: there seems to be a middle setting where magic is supposed to happen in addition to parent-controlled white and blacklists [08:47:01.0000] othermaciej: unlike IE, Safari doesn't say what the magic is [10:47:00.0000] http://forums.whatwg.org/viewtopic.php?t=151 [11:06:00.0000] I don't think introducing even more entities is a good idea [11:17:00.0000] so I'm trying to come up with the most reasonable/portable way to represent this data as HTML: http://h3h.net/images/data-representation-01.png [11:18:00.0000] I think a DL matches somewhat, so this would be good: http://pastie.caboo.se/153841 [11:18:01.0000] but I can't think of a way to style that reliably [11:19:00.0000] if you care about portability, I think you lose; I'm not sure how you'd work with that other than with ~ or + selectors [11:19:01.0000] right. even then, separating the dt/dd+ blocks from one another is near impossible [11:19:02.0000] so I guess each data block has to be its own table or dl [11:20:00.0000] supremely disappointing [11:21:00.0000]

Members

stuff [
more stuff]
or something [11:21:01.0000] yeah, but then I have the same sibling grouping problem [11:22:00.0000] h3h, looks like a job for nested tables :-) [11:22:01.0000] tables really don't work because they assume a specific row/column layout [11:22:02.0000] .outer > .inner { } .outer > .header { } ? [11:22:03.0000] in this case I just have header:data groupings [11:22:04.0000] which could be displayed vertically, horizontally, or otherwise [11:23:00.0000] then dl/dt/dd might be appropriate, though may be difficult to achieve the intended style [11:23:01.0000] so it would have to be more like

Header

data

Header 2

data 2

...
[11:23:02.0000] which is disgusting, frankly [11:24:00.0000] yeah, I think I'm going to have to go with a new DL for each one [11:24:01.0000] it's web development :-) [11:24:02.0000] less terrible than the alternatives [11:25:00.0000] it's like I want the semantics of a "table row" (which can contain a header and one or more data elements) without the notion of an actual row [11:26:00.0000] the simplest table markup would be perfect if I could style it vertically [11:27:00.0000] table-layout: transposed; or something :) [11:30:00.0000] actually, I might be able to pull that off... [11:40:00.0000] hah: http://paste.css-standards.org/32610/view [11:40:01.0000] though it displays upside down in Safari [11:42:00.0000] and sideways in IE [11:42:01.0000] Firefox and Opera get the prize for simplest reverse engineering of default styles [11:56:00.0000] h3h: try making it strict [11:56:01.0000] hm good point [11:57:00.0000] ah nice shooting. works fine with strict [11:58:00.0000] in Safari [11:58:01.0000] IE still sucks of course [11:58:02.0000] Would you have it any other way? [11:58:03.0000] :( [12:23:00.0000] othermaciej lives! [12:23:01.0000] hello Hixie [12:23:02.0000] I've been on vacation the past week [12:23:03.0000] back now [12:23:04.0000] so i heard [12:24:00.0000] you might be interested in http://hixie.ch/specs/dom/messages/0.9 and http://hixie.ch/specs/dom/workers/0.9 [12:24:01.0000] once you've dealt with your e-mail [12:24:02.0000] Hixie: are you serious about "The eyes go orange if you view the test zoomed in or zoomed out: This is a bug."? [12:24:03.0000] see https://bugzilla.mozilla.org/show_bug.cgi?id=418235#c6 [12:25:00.0000] expecting that transparency offset thing to work when zoomed seems a bit much [12:29:00.0000] so Hixie's saying that image scaling algorithms that use any kind of interpolation are not allowed? [12:31:00.0000] it's certainly suboptimal [12:32:00.0000] one could imagine a perfect UA that rendered all the graphics at 1:1, then scaled the composited graphic down [12:32:01.0000] and then overlapped the text on top of that [12:32:02.0000] i'm not saying that's necessarily even remotely easy or likely to ever be implemented [12:33:00.0000] actually that is feasible, it's full-screen antialiasing [12:33:01.0000] but it [12:33:02.0000] 's expensive [12:33:03.0000] i'm sure [12:34:00.0000] but it's hard to argue that it's not what is intended by the author :-) [12:34:01.0000] hopefully this issue doesn't come up much in real-world web pages :) [12:34:02.0000] probably comes up more than you'd like, and less than makes it worth it :-) [12:35:00.0000] well, in this case we might wilfully ignore the author's intent since interpolation produces good results for real Web pages [12:35:01.0000] as in, I haven't seen a real bug filed related to this [12:36:00.0000] well we haven't shipped zoom yet :-) [12:37:00.0000] true, and IIRC only Mac builds use interpolation [12:37:01.0000] but still, there are something like 300K active beta3 users [12:39:00.0000] rue [12:39:01.0000] true [12:41:00.0000] like i said, i doubt that it's worth fixing [12:41:01.0000] bbl [13:54:00.0000] hmm. now internationalization and accessibility & ethics are at odds: http://lists.w3.org/Archives/Public/www-archive/2008Feb/0030.html [13:56:00.0000] i've given up reading through his replies [13:58:00.0000] /me wonders how fast his note about will cause a stirr [13:59:00.0000] /me wonders if annevk did it on purpose [13:59:01.0000] well, I did want to point out that his point of view has caused issues in the past, but I didn't want to say too much [14:00:00.0000] I think it's rather well put, but probably only if you agree with my line of thought [14:03:00.0000] annevk: well, gotta love a situation where one way leads to getting accused of breaking accessibility and the other would lead to accusations of breaking internationalization [14:05:00.0000] It's hard for me to see what his real issue is. That some browsers are still broken? [14:06:00.0000] annevk: so I gather from his feedback [14:06:01.0000] ok, so I did read the important parts [14:06:02.0000] of course, it is well-known that there are older versions of IE still out there that don't support IDN [14:06:03.0000] but IE7 supports IDN, right? [14:06:04.0000] guess first sentence/last sentence, first paragraph/last paragraph, might be applicable to e-mail too [14:06:05.0000] hsivonen, yeah [14:07:00.0000] they also experimented with IDN for e-mail I believe [14:07:01.0000] annevk: I think you're probably right about aria-* and , at least from the pov of author confusion. I can't imagine any spec that forces us to deal with bar is going to be easy for authors to understand [14:08:00.0000] "well, if the OBJECT model failed, it was due to lousy implementation decisions" [14:08:01.0000] Ouch: "Some weeks ago you quoted an ISO standard I haven't heard of before for your definition of "valid". If that ISO standard has "congruent de facto guidance" in its definition trash it or maybe put it where you have DIS 29500." [14:08:02.0000] jgraham: I happen to have a use case for having an input labeled by a there though or three separate forms [14:10:00.0000] i guess both approaches are ok somehow [14:11:00.0000] hsivonen: I don't think all the features are bad, just that making it all easy to understand will be tough [14:12:00.0000] pretty clear that HD DVD is dead now [14:12:01.0000] glad I didn't buy this Xbox add-on [14:42:00.0000] hsivonen: do I remember hearing that you have some doctype parsing tests? 2008-02-19 [22:30:00.0000] man, people have go to start realising that when an implementor says he's not doing something, arguing with him is a bad idea [22:54:00.0000] Hixie: do you know of any doctype parsing tests? [23:55:00.0000] Hixie, which issue are you referring to, and who's arguing with them? [23:56:00.0000] appformats list [23:56:01.0000] or maybe webapi [23:59:00.0000] haven't read appformats in a while, not too sure what's going on there [00:01:00.0000] wow, another ala on version targeting [00:01:01.0000] I uploaded my disposistion of comments last night http://dev.w3.org/2006/webapi/selectors-api/disposition-of-comments [00:01:02.0000] Yeah :/ [00:03:00.0000] oh well, ala hasn't been interesting for a long time now [00:03:01.0000] aroben: I'm known to have http://hsivonen.iki.fi/doctype/ but not test cases for parsing broken doctypes [00:03:02.0000] hsivonen: ah, thanks [00:04:00.0000] I think html5lib might have some [00:04:01.0000] hmm. I still never wrote anything about the version thingy this time round. maybe I should, if this ALA article says nothing but misinformation about it (again) [00:05:00.0000] annevk: will look [00:05:01.0000] aroben, http://html5lib.googlecode.com/svn/trunk/testdata/tokenizer/test3.test [00:05:02.0000] http://html5lib.googlecode.com/svn/trunk/testdata/tokenizer/test2.test [00:06:00.0000] http://html5lib.googlecode.com/svn/trunk/testdata/tokenizer/test4.test [00:06:01.0000] http://html5lib.googlecode.com/svn/trunk/testdata/tokenizer/test1.test [00:06:02.0000] unfortunately not grouped [00:06:03.0000] annevk: still very helpful [00:10:00.0000] annevk: are the html5lib tests up to date with the recent change regarding '>'? [00:10:01.0000] no :( [00:15:00.0000] at least, not as far as I know [00:17:00.0000] actually, yes, the tests are up to date thanks to Philip` it seems [00:28:00.0000] annevk: no new stuff in the new ALA [00:29:00.0000] boring [00:30:00.0000] can someone tell me the result of http://tc.labs.opera.com/css/namespaces/ 002 and 003 in Safari? [00:31:00.0000] annevk: in a recent nightly both of them have a green background [00:32:00.0000] thanks [00:32:01.0000] In 3.0.4 as well [00:32:02.0000] does Safari pass 001? [00:32:03.0000] Yes [00:52:00.0000] I don't really understand Boris' last comment: "but it wouldn't address the issue of what implementations are allowed to actually do to stop misbehaving NSResolvers" [00:53:00.0000] yeah, what's a misbehaving nsresolver? [00:53:01.0000] i mean, you could of course do all kinds of quirky things in script, but that's not restricted to NSResolver [00:53:02.0000] how does resolving all namespaces prior to traversing the DOM, thus letting the NSResolver do anything it likes beforehand, get in the way of actually finding matches afterwards? [00:54:00.0000] /me will respond [01:00:00.0000] "Invited Expert will participate in the W3C Group in a decent way." [01:04:00.0000] A year by already [01:08:00.0000] oh [01:08:01.0000] /me thought the HTML WG started in March [01:09:00.0000] A year minus 1 month then :) [01:09:01.0000] Have to reinvite myself as an expert.. [01:11:00.0000] i wonder how many people won't do that [01:13:00.0000] I think you get bugged with a mail each week, until you say yes or no [01:14:00.0000] Pretty irritating, so saying "yes, I'm still teh expert you need!11" does the trick :) [01:15:00.0000] krijnh: do (public) Invited Experts have to renew their participation annually or what is this about? [01:15:01.0000] Yeah, I guess so [01:15:02.0000] And there's a new EULA or something [01:16:00.0000] krijnh: what's the diff in the "EULA"? [01:16:01.0000] http://www.w3.org/Consortium/Legal/2007/06-invited-expert.html [01:16:02.0000] No idea, I'm not an invited expert on EULAs :p [01:17:00.0000] Prolly just a clarification for the 2002 version [01:17:01.0000] With new elements and a new API [01:18:00.0000] Can't they just create a new group for this? Invited Wannabe Experts or something.. [01:18:01.0000] hmm. I don't like it that the W3C wants to use copyright to prevent spec branching [01:20:00.0000] me neither [01:20:01.0000] or for test suite branching for that matter [01:20:02.0000] it's silly [01:20:03.0000] the freedom to fork a Free Software project is an important deterrent against the main project going crazy [01:20:04.0000] should be likewise for specs [01:21:00.0000] I'm glad that's the case for the WHATWG's HTML5 spec [01:21:01.0000] maybe they realize they're going crazy in some twisted way [01:43:00.0000] Hixie, how do you suggest resolving the hostile-NSResovler issue, if Anne's suggestion is no good? [01:49:00.0000] Lachy: I'm not Hixie, but it seems to me that the way to deal with NSResolver side effects in an interoperable way is to spec when exactly NSResolver is called into [01:50:00.0000] no, we don't want to go there [01:51:00.0000] that would not allow optimizations [01:51:01.0000] hmm. how about requiring the script to push a mapping hash table to the browser ahead of time so that no NSResolver is needed? [01:51:02.0000] *[foo|x]:not([foo|x]) for instance [01:52:00.0000] annevk: you could put the optimization into the spec itself [01:52:01.0000] that's difficult [01:52:02.0000] what if someone finds an even better optimisation later, and the spec would then prevent them from implementing it [01:53:00.0000] it would constrain implementations and makes it harder to introduce new selectors [01:53:01.0000] is there a reason why prefix resolution needs to be dynamic instead of a static mapping that is given the the browser at the time of the API call? [01:54:00.0000] hsivonen, I think it was because the resolver was based upon the design of the XPathNSResolver [01:55:00.0000] annevk: would implementation be constrained too much if implementation were required to [01:55:01.0000] 1) parse the selector [01:55:02.0000] crap. meeting time :-( [01:55:03.0000] 2) resolve each prefix from left to right exactly once [01:55:04.0000] 3) run the selector [01:55:05.0000] ? [01:56:00.0000] for the case I gave above you don't need to resolve the prefix in theory [01:56:01.0000] annevk: ah right. but are such pathological edge cases worth optimizing [01:57:00.0000] annevk: they don't serve a useful purpose for authors [01:57:01.0000] people say yes [01:57:02.0000] i'm not sure [01:57:03.0000] (people being bjoern and hixie, so maybe...) [01:57:04.0000] :somenewfeature(foo|x) /* CSS4 browsers need to resolve this prefix, but CSS3 browsers can't know they're meant to */ [01:57:05.0000] that would not parse Philip` [01:58:00.0000] and throw a SYNTAX_ERR [01:58:01.0000] Ah [02:02:00.0000] as far as I can tell, this problem is mostly the same problem that was pointed out when Dave Raggett's XForms Transitional was discussed. [02:03:00.0000] and really the way to make sure that the side effects are the same in all browsers is to clamp down what the calls to the code with the potential side effects are [02:03:01.0000] no [02:04:00.0000] no? [02:04:01.0000] the issues is not about all browsers doing the same [02:04:02.0000] it doesn't have to be fully deterministic [02:04:03.0000] oh. what is it then? [02:04:04.0000] the issue is more what to do in the case that the authors makes a NSResolver that halts the browser in some way [02:05:00.0000] how has that issue been solved for treewalker node filters? [02:05:01.0000] i don't think it has been [02:06:00.0000] Well, how would halting the browser in a NSResolver be any different from halting the browser otherwise? [02:06:01.0000] Gecko running scripts in the UI thread is an issue [02:06:02.0000] for the case of a hanging NSResolver, I assumed that would be dealt with the same way all hanging scripts in the browser, by eventually timing out and halting execution completely [02:07:00.0000] bad scripts would be less harmful if each browsing context ran its layout and scripts on a private thread [02:08:00.0000] you don't need threading for that [02:09:00.0000] /me wonders what happens if you use "yield" in an NSResolver [02:09:01.0000] annevk: without threading you need code to reach a timeout checkpoint often enough [02:09:02.0000] what does yield do? [02:10:00.0000] Oh, actually, I suppose it wouldn't do anything very interesting at all [02:10:01.0000] It returns, but preserves internal state for the next call [02:10:02.0000] Lachy: http://developer.mozilla.org/en/docs/New_in_JavaScript_1.7#Generators_and_iterators [02:13:00.0000] I was thinking of something that preserves more than one stack frame, but that doesn't exist in JS so that's okay [02:13:01.0000] hsivonen, code gets icky, yeah [02:17:00.0000] Even if you run code in separate threads, you need a way to cleanly terminate a thread when the user's bored with it sitting at 100% CPU, which makes code a bit icky too [02:18:00.0000] Except actually that's maybe only a tiny bit and you wouldn't even notice it [02:24:00.0000] Philip`: well, yeah, you'd need a mechanism that makes sure that the killed thread isn't holding any monitors/semaphores/etc. that would be left in an inconsistent state [02:24:01.0000] grmbl, my "we can't" argument is a real argument why we can't do
magic, but the overall reason is of course that it's silly and would be quite difficult, performance hit, etc. [02:25:00.0000] Philip`: which isn't exactly trivial unless there are only a couple well-defined places where locking can happen [02:26:00.0000] (BTW, this is the reason I haven't implemented Hixie's suggestion to limit the time a validation transaction is allowed to take instead of limiting the number of bytes Validator.nu is willing to ingest) [02:46:00.0000] Hmm. Jonas' suggestion to mention the hanging nsresolver case in the security section is a good idea... Oh, wait. It already is! [04:17:00.0000] fwiw, in case anyone cares about arguing against a new HTTP verb for the ping feature: AJP 1.3 seems to assume a closed list of HTTP verbs, so minting a new one would have a non-trivial cost [04:19:00.0000] what's AJP? [04:20:00.0000] Lachy: it's a protocol that Apache speaks with Java servlet containers [04:21:00.0000] ok, so it's something that would potentially be used on a server for dealing with pings. [04:22:00.0000] however, if you consider that adding a new HTTP verb can't be that much effort, and that any system to be set up for handling pings in the future could easily upgrade. It doesn't require all servers to be upgraded. [04:22:01.0000] the bigger problem with ping is getting client support [04:23:00.0000] Lachy: unless I'm misreading the source of an AJP 1.3 implementation, it seems to me that AJP 1.3 communicates the HTTP verb as a single byte whose meaning has to be agreed to by both ends of the AJP pipe [04:23:01.0000] oh, in that case, it might be a little more difficult [04:24:00.0000] It just needs a single centralised service which understands the PING requests, and then every web developer can make use of that service to collect data and to retrieve summaries and reports [04:24:01.0000] Lachy: I'm not saying it is good design. I'm just saying that there's a common deployment setup with which going beyond the WebDAV set of verbs is going to be expensive [04:24:02.0000] It's pointless for people to run the ping-processing software on their own servers [04:27:00.0000] Philip`, I don't think it's entirely pointless, but having a single centralised service might work for some system [04:28:00.0000] e.g. google analytics, which currently uses JS to notify the google server of certain events, could make use of ping in that way [04:30:00.0000] http://tomcat.apache.org/connectors-doc/ajp/ajpv13a.html#method [04:31:00.0000] given that webdav is still expanding i assume they'll eventually reconsider their api [04:31:01.0000] note that is says "Later version of ajp13, when used with mod_jk2, will transport additional methods, even if they are not in this list." but mod_jk2 is dead and mod_jk is what people are supposed to be using again [04:33:00.0000] annevk: perhaps. I'm just pointing out that considering the requirements Hixie had for the access-control implementability, a new HTTP verb is on a new level of difficulty [04:34:00.0000] fwiw, the mechanism for encoding headers in AJP13 (I guess it's really 13 and not 1.3) is extensible [04:34:01.0000] I think the failure to use the same mechanism for the verb is a design bug, but software has shipped [05:25:00.0000] when do we get the ALA article about version targeting being all wrong? [05:25:01.0000] or is it supposed to be one-sided? [05:30:00.0000] How would you go about changing jeffrey Z's mind? [05:31:00.0000] annevk: http://www.alistapart.com/articles/theyshootbrowsers sounds non-positive about it [05:32:00.0000] itpastorn, i'd put him on the QA team of non-IE browser for a year or two [05:32:01.0000] Philip`, I was going by the summary on zeldman.com [05:32:02.0000] itpastorn: Hypnosis [05:32:03.0000] (since that requires less effort than thinking of logical arguments) [05:37:00.0000] OK, so maybe we will get JZ and Eric Meyer to change their minds. Now, how would we get the MSIE team onboard? I seriously can't think of any way to change their minds. [05:38:00.0000] As I see it now it is an issue of damage control [05:38:01.0000] Change their minds before releasing IE8? [05:39:00.0000] (Perhaps it's more possible to make their IE8 experience cause them to reconsider for IE9) [05:39:01.0000] ((like how IE7 made them reconsider IE8)) [05:40:00.0000] Yes, the ball-and-chain effect of supporting multiple rendering modes will sooner or later take its toll [06:27:00.0000] annevk: presumable setting a header ahead of GET could be specified to trigger pre-flight? [06:27:01.0000] s/ble/bly/ [06:39:00.0000] hsivonen, that's quite a neat idea [06:52:00.0000] hsivonen, I put it on the list [07:17:00.0000] annevk: do you know if there are any test cases for cssom .styleSheets in combination with @import? [07:18:00.0000] i thought .styleSheets was not supposed to be populated with that [07:21:00.0000] zcorpan, I don't really understand what you're asking for [07:23:00.0000] sorry, i meant .styleSheet (on CSSImportRule) [07:24:00.0000] no, feel free to add to http://tc.labs.opera.com/apis/cssom/ [07:24:01.0000] ok [07:28:00.0000] no directory structure? [07:28:01.0000] feel free to do that for your tests [07:29:00.0000] i was making tests to see how I should spec stuff and needed a place to dump them [07:35:00.0000] ok [07:35:01.0000] i'm not sure if it was the right approach [08:11:00.0000] annevk: is .ownerNode null for imported style sheets? [08:25:00.0000] yes [08:31:00.0000] thanks [08:31:01.0000] (that was not super clear to me reading the spec) [08:34:00.0000] suggestions welcome [08:34:01.0000] i'm not really sure how to phrase the conformance stuff for all that yet [08:45:00.0000] can it be null in other cases? [08:56:00.0000] HTTP Link: [08:56:01.0000] if we're keeping that [08:57:00.0000] sorry [08:57:01.0000] in the case it would not be null, duh [08:58:00.0000] It wouldn't be especially equiv to HTTP if it differed in that way [08:59:00.0000] it's just a name [10:36:00.0000] i tried getting an account and it failed [10:36:01.0000] if anyone has a wordpress.com account [10:36:02.0000] it would be cool to ask this guy what it is he would actually like: [10:36:03.0000] http://mrpointy.wordpress.com/2008/02/19/ding-dong-the-frame-is-dead/ [10:47:00.0000] wow. I can actually remember my user/pass combo. on just the second attempt [10:48:00.0000] Hixie: + normal invitation to participate? :P [10:51:00.0000] comment in moderation queue [11:00:00.0000] gsnedders: sure [11:01:00.0000] dbaron: is there anything anywhere online about your table/desk based implementation of HTML? [11:01:01.0000] I couldn't find anything googling for it [11:01:02.0000] no [11:01:03.0000] not that I know of, anyway [11:02:00.0000] well, there are logs of here, but that isn't an overly convincing place to cite :) [11:48:00.0000] Hixie: is there anything in acid3 that assures a tree DOM is being used, and not something like IE's? [11:50:00.0000] no, because html4 doesn't define how you handle the parse errors that get you a non-tree DOM [11:53:00.0000] ergh. [12:33:00.0000] /me looks through the X-UA-Compatible comments and sighs [12:33:01.0000] "I code all my sites to be xhtml strict and I expect a browser to render them that way" — yes, because "xhtml strict" specifies rendering [12:48:00.0000] such comments indicate an incomplete understanding of the web, but their sentiment is well placed and i agree with it [12:54:00.0000] Hixie: btw, did you see my results from validating the Happy Cog portfolio? [12:55:00.0000] yeah, briefly. it ended up in one of my folders. [12:56:00.0000] ok [14:30:00.0000] Philip`: do you have any useful information on the cite="" attribute of q and blockquote elements? [14:37:00.0000] Hixie: Only that it's quite rare [14:37:01.0000] In 16K pages, I found it only on http://www.kentuckyinjurylawblog.com [14:38:00.0000] and on another 8K (at http://canvex.lazyilluminati.com/survey/2007-07-17/analyse.cgi/attr/cite ) I didn't find it at all [14:39:00.0000] http://canvex.lazyilluminati.com/survey/2007-07-17/analyse.cgi/tag/q - hmm, nobody uses anyway [14:39:01.0000] so on your 16k pages, that's 100% of uses used it correctly! :-) [14:39:02.0000] http://canvex.lazyilluminati.com/survey/2007-07-17/analyse.cgi/tag/blockquote - blockquote much more popular, but very rarely any attributes [14:41:00.0000] Smells like indentation :) [14:41:01.0000] yeah [14:43:00.0000] what's that like, the smell of presentational markup? [14:44:00.0000] rather good looking, but crappy food [14:45:00.0000] I've never used blockquote with attributes, but always for quotations. Is that so wrong? [14:45:01.0000] no, that's fine [14:47:00.0000] /me has cite attributes automatically created in quoted replies on a message board system he wrote. [14:50:00.0000] /me wonders if would generally be better than @cite [14:53:00.0000] how about implicit rel=cite [14:53:01.0000] paragraph scoped or some such [14:54:00.0000] ? In what sense implicit? [14:55:00.0000]

... ...

[14:56:00.0000] is cite for [14:57:00.0000] That seems like it could get messy fast

When I went to the theatre, I was reminded of the shakespeare quote All the world's a stage

[14:59:00.0000] I guess doesn't work so well for replacing @cite because there's no way to attach the attribution to the quotation and it's usually the attribution that would be linked [15:00:00.0000] labeledby! [15:00:01.0000] or labelledby [15:00:02.0000] no comment :) [15:01:00.0000] p contentof div [15:01:01.0000] it seems both spellings are used by the accessibility community [15:01:02.0000] awesome [15:01:03.0000] Then you don't have to worry about where you place your stuff [15:01:04.0000] http://www.google.com/search?q=labeledby+w3c [15:01:05.0000] http://www.google.com/search?q=labelledby+w3c [15:02:00.0000] aria ftw! [15:02:01.0000] labeledby sameas labelledby [15:03:00.0000] Dashiva: Isn't contentof spelled "owns" [15:04:00.0000] Not sure. Just offhand, I'd assume a div could own an element that wasn't part of its content. [15:05:00.0000] Did you mean: pwns [15:05:01.0000] Dashiva: I think aria allows you to do
This is my div

My div is somehow a child of this

[15:05:02.0000] annevk: :-) [15:38:00.0000] ok my plan is to keep cite="" basically unchanged from html4 [15:53:00.0000] can anyone find a word that can be used to describe the category to which the following things belong?: a book, a paper, a song, a movie, a TV show, a painting [15:54:00.0000] ...but that does not include: a person, a boat [15:54:01.0000] media? [15:55:00.0000] sadly "media" is even more overloaded that the word i've been using so far ("work") [15:58:00.0000] "a work" is relatively accurate [15:58:01.0000] particularly if backed by that list of examples [15:58:02.0000] yeah, that's all i could find so far [15:58:03.0000] Should it include things like plays, where it's not a written or recorded work that can be directly referenced? [15:59:00.0000] /me adds theatre productions, plays, operas, musicals to the list [15:59:01.0000] Most plays are written, no? But there are obviously other examples of oral traditions 2008-02-20 [16:08:00.0000] other oral traditions are also "works", though a bit more loosely... [16:09:00.0000] Hixie, "creative work"? [16:10:00.0000] has Microsoft announced anything in particular about better DOM/JS support in IE8? [16:10:01.0000] Papers aren't really creative works [16:10:02.0000] Zeldman keeps mentioning it but I don't see an official announcement or anything [16:10:03.0000] mpt: hm, that might work [16:10:04.0000] oh yeah, creative fails for things like research papers [16:10:05.0000] (cue jokes) [16:11:00.0000] Why does it? [16:11:01.0000] but seriously, i think creative would help as much as it hurts here [16:11:02.0000] because people associate "creative" with art [16:11:03.0000] rather than research [16:11:04.0000] Seriously, papers are creative works [16:11:05.0000] A more awkward example might be the telephone directory [16:12:00.0000] (although I agree that many people wouldn't use the term in that context, I would strongly disagree with them) [16:16:00.0000] i'll just stick to "work" with a long list of examples [16:16:01.0000] and an explicit ban on names of people or ships [16:16:02.0000] This is for the element, right? [16:16:03.0000] yep [16:17:00.0000] /me still doesn't think it's very useful [16:17:01.0000] me either [16:17:02.0000] but it's used more than [16:17:03.0000] and you should have seen the hell that the xhtml2 wg received for trying to drop it [16:18:00.0000] Yeah, I accept that people want to use it even though doing so has no obvious benefits [16:19:00.0000] that was because Mark Pilgrim actually used it for something [16:19:01.0000] that's no longer the case [16:20:00.0000] i've already spoken to mark about this [16:20:01.0000] (and i'm not saying that the aforementioned wrath is a reason to keep it, i'm just mentioning it) [16:21:00.0000] for ship names though? [16:21:01.0000] I guess doing something interesting with citations will have to be left to the microformats people [16:22:00.0000] i guess since it renders in italic it's ok :) [16:22:01.0000] i said ship names were explicitly banned :-P [16:22:02.0000] use [16:22:03.0000] (which is btw one of the reasons why Anne said: ... doesn't make much sense; "Anne" there typically wouldn't be italicized) [16:22:04.0000] oops [16:23:00.0000] ah, people too [16:31:00.0000] wow, i didn't realise that the first REC the w3c ever did was PNG [16:33:00.0000] i love the comment on libpng.org: [16:33:01.0000] "(By the way, despite the implications in some of CompuServe's old press releases and in occasional trade-press articles, PNG's development was not instigated by either CompuServe or the World Wide Web Consortium, nor was it led by them. Individuals from both organizations contributed to the effort, but the PNG development group exists as a separate, Internet-based entity.)" [16:33:02.0000] i wonder if we should publish something like that on the whatwg page :-P [16:50:00.0000] http://www.whatwg.org/specs/web-apps/current-work/#the-cite [16:53:00.0000] "even if people call that person a piece of work" :p [16:56:00.0000] i'd like HTML5 to define this: "window.name = '1'; name = '2'; w(name); w(window.name)" [16:57:00.0000] there's some magic going on there [16:57:01.0000] is that the replaceable stuff? [16:57:02.0000] well, it seems that some browsers are able to distinguish between window.name and name (both in the global scope) [16:58:00.0000] i think that's different from replaceable, but I'm not sure [16:59:00.0000] odd [16:59:01.0000] what's not odd is that sites rely on this [16:59:02.0000] when you say "browsers" [16:59:03.0000] which browsers? [16:59:04.0000] firefox doesn't seem to distinguish them [16:59:05.0000] Firefox and Internet Explorer [17:00:00.0000] http://software.hixie.ch/utilities/js/live-dom-viewer/?%3Cscript%3Ewindow.name%20%3D%20%271%27%3B%20name%20%3D%20%272%27%3B%20w(name)%3B%20w(window.name)%3C%2Fscript%3E [17:00:01.0000] window.name = 'a'; name = 'b'; name + ' ' + window.name; on http://software.hixie.ch/utilities/js/js-eval-window/ evaluates to "b b" [17:00:02.0000] I get 2, 1 in the log [17:00:03.0000] i get 2,2 [17:01:00.0000] Mozilla/5.0 (X11; U; Linux i686 (x86_64); en-US; rv:1.9b2pre) Gecko/2007112904 Minefield/3.0b2pre [17:01:01.0000] Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9b4pre) Gecko/2008021204 Minefield/3.0b4pre [17:01:02.0000] wow, mine is old [17:01:03.0000] get "b a" for yours [17:03:00.0000] ff2 gives b b/2 2 too [17:04:00.0000] yeah [17:04:01.0000] ff3 nightly crashes on startup [17:04:02.0000] /me nukes his profile [17:05:00.0000] heh, I even get 2 2 in IE7 [17:05:01.0000] maybe I'm testing this wrongly [17:05:02.0000] could be that it only applies to nested browsing contexts or so, I guess :( [17:07:00.0000] Hixie, btw, could you add "html" and "css" as resources to the live dom viewer? [17:08:00.0000] maybe even script that does w('run') or something like that [17:11:00.0000] /me -> bed [17:12:00.0000] adbded [17:12:01.0000] added even [18:37:00.0000] othermaciej: the workers idea i proposed included a DOMImplementation object, from which you can create a Document [18:37:01.0000] afk bbiab [18:38:00.0000] Hixie: I haven't read over all the proposals closely yet, but that would indeed raise the same issue as the X part of XHR being present [18:39:00.0000] I can see how it could be useful but at least in WebKit I think it would significantly increase the implementation complexity [19:18:00.0000] eek yes that would be real pain [19:43:00.0000] really? [19:43:01.0000] huh [19:44:00.0000] i would have thought you could run a DOM on a different thread without problems, what of the DOM implementation is not thread safe? [19:49:00.0000] depends on the implementation, but it's easy to imagine that the DOM uses global internal caches [19:50:00.0000] and it's not just the DOM but everything the DOM touches [19:50:01.0000] canvas, video, you name it [19:51:00.0000] true [19:52:00.0000] it would be sad if it was not available [19:52:01.0000] creating markup is one of the things i'd expect to happen in a worker [19:52:02.0000] even if it's not manipulating forms, videos, audio, etc. [19:53:00.0000] strings and innerHTML baby [19:53:01.0000] gotta save something for HTML6 [19:53:02.0000] creating markup with strings is sub-optimal [19:53:03.0000] but maybe e4x can find a use finally :-) [19:53:04.0000] The JS library writers say it's actually optimal in practice :-) [19:53:05.0000] i didn't mean performance-wise [19:53:06.0000] :-) [19:56:00.0000] bbl [01:02:00.0000] http://blog.mozilla.com/rob-sayre/2008/02/19/bloaty-parts-of-the-whatwg-html5-specification-that-should-be-removed/ [01:05:00.0000] http://diveintomark.org/archives/2008/02/19/all-these-years [01:06:00.0000] Does he know there's a element as well now? :) [01:37:00.0000] "they’re taking away my right to cite a person" [01:52:00.0000] Dashiva: Yeah. I managed to restrain myself from asking which set of laws guaranteed that particular right [01:53:00.0000] /me doesn't remember it being part of the EU convention on human rights [01:56:00.0000] A right to life, a right to liberty and security, a right to freedom of thought, conscience and religion, a right to use for people's names? [01:56:01.0000] Doesn't quite fit, I think [01:56:02.0000] Maybe it fits into the religion part [02:07:00.0000] here's some more from that stuff yesterday: http://wiki.mozilla.org/Gecko:SplitWindow [02:08:00.0000] i guess it's unlikely to affect the topmost browsing context [04:05:00.0000] damn, should be allowed for people's names. How else is one expected to markup something like:

John Smith said Hello World!

? [04:05:01.0000] I use it all the time for people's names when I quote from blog entries. I usually use it for both the author and the article title. [04:06:00.0000] What benefit do people get from you using it like that? [04:07:00.0000] I don't know, it's what I do [04:08:00.0000] maybe I've been using it wrongly all these years. I don't know [04:08:01.0000] So it's superstition? :-) [04:09:00.0000] i did that too, but it recently learned that the convention is that those names should not be in italic and other than that there was no apparent benefit [04:10:00.0000] ok. If there's no typographical convention for people's names, then I guess I'm wrong [04:12:00.0000] s/but it/but I/ ... [04:13:00.0000] Lachy: in Finland, there's a convention in newspapers and magazines (not books) to bold the first instance of each personal name in an article [04:13:01.0000] Lachy: clearly, not what is good for [04:14:00.0000] [04:15:00.0000] yes [04:27:00.0000] lol, someone wants to put advertisements on a few of my pages for inifinite time for a one-time 200 USD [04:28:00.0000] annevk, I get spam like that occasionally (though usually caught by my spam filter) [04:28:01.0000] it's a serious offer [04:28:02.0000] annevk: was the price disclosed in the initial offer? the people who want to advertise on my site never disclose the money figures up front but promise to make it worthwhile for me [04:28:03.0000] yeah, I even had some serious offers once [04:29:00.0000] It prompted me to publish my advertising policy on my site [04:29:01.0000] hsivonen, oh no, I replied [04:29:02.0000] http://lachy.id.au/about/advertising [04:29:03.0000] i do run advertising [04:30:00.0000] but it's finite time, has limited impact on my software, has zero maintenance cost, and gives me a share that i think is acceptable [04:30:01.0000] on annevankesteren.nl? [04:30:02.0000] I've never seen any ads [04:30:03.0000] it's hardly noticable [04:31:00.0000] search for "beslist.nl" [04:31:01.0000] (inline search) [04:31:02.0000] oh [04:31:03.0000] that's the best advertising I've seen. all sites should do that! :-) [04:32:00.0000] hmm. seems like they are partly paying for the google juice [04:33:00.0000] google would say those links should have nofollow on them :) [04:33:01.0000] these advertisers seem to think they are paying for google juice but aren't getting any: http://www.w3.org/Consortium/sup [04:35:00.0000] hsivonen, true [04:36:00.0000] could be debated [04:37:00.0000] hsivonen, heh, probably the only w3.org page with "Generic Viagra" [04:38:00.0000] or maybe not, check this out: http://www.google.com/search?q=site:w3.org+viagra [04:38:01.0000] have Google stated that links on pages like that pass no link pop? [04:38:02.0000] just ebcause they are told not to follow them, doens't mean they can't take them into account for link pop [04:39:00.0000] Camaban: IIRC, nofollow is a misnomer and means nopagerank [04:40:00.0000] well, when I first mentioned nofollow, I was on about rel=nofollow, and yes, google suggested using that to combat spam, and have since recommended it for marking up ads as well [04:40:01.0000] because they don't feel those links should pass PR [04:41:00.0000] but the robots meta tag is a bit different [04:41:01.0000] oh [04:41:02.0000] the W3C seems to think they aren't selling google juice, though [04:42:00.0000] heh, well it's come up in SEO circles a few timesa about the 'paid links' the W3C has on that page [04:42:01.0000] but it's pretty obvious that those who buy placement on that page think they are buying it [04:42:02.0000] especially as it's a high authority domain [04:43:00.0000] now and then i read something or hear something from these so-called SEO's and i often doubt they know what they're talking about [04:43:01.0000] toolbar suggest that particular page has a PR of 8, a link from like that is worth a fair bit :) [04:44:00.0000] annevk: there are lots of people who call themselves SEO guru's who think meta keywords still make a difference in ranking [04:44:01.0000] on the other hand, there are numerous people who do have a clue :) [04:55:00.0000] ok, having checked, it seems Matt Cutts has stated that meta nofollow does the same as rel=nofollow, so those links should indeed NOT be passing google juice [04:56:00.0000] the page is indexed, so the links can be found as part of the page, but they aren't benefiting the recipients in terms of link popularity :) [07:59:00.0000] Philip`: ping [08:01:00.0000] http://philip.html5.org/data/cite.txt in case anyone is interested [08:01:01.0000] zcorpan: Hello [08:02:00.0000] (Is there some WYSIWYG editor that encourages people to use when they want presentational italics?) [08:03:00.0000] knowing the general quality of WYSIWYG editors, probably [08:03:01.0000] that's a lot of support for CITE meaning citation not just title of work. [08:06:00.0000] Philip`: (nothing widely used, at least) [08:07:00.0000] /me is just wondering how No.. No....please, no electric shocks. comes about [08:08:00.0000] Actually, I guess a WYSIWYG editor usually wouldn't generate misnested tags [08:08:01.0000] from http://achristiancounselor.com/witness.html ? [08:09:00.0000] Yes [08:13:00.0000] Philip`: OpenOffice.org and derivatives produce for what is labeled as Quotation in the English UI [08:14:00.0000] and that matches the use on that page [08:14:01.0000] that's not an uncommon confusion [08:14:02.0000] /me wonders if that's been filed as a bug [08:14:03.0000] Hmm, that doesn't match the other uses on that page [08:15:00.0000] Quotation is Zitat is German and OOo was originally developed by a German team [08:15:01.0000] (Jesus is the same, yesterday, today and forever. He healed then, so He heals now. That was as clear as a bell and simple as pie.) [08:15:02.0000] Philip`: it seems that two tests in the canvas test suite have no reference image, yet. when you have compared the results, have you skipped them or set them to pass or fail? [08:16:00.0000] Philip`: It's possible that having inserted a quotation the author then copied and pasted to get the same formatting or something. [08:17:00.0000] zcorpan: I thought there were more than that with no reference image, but only in cases where the result is determined automatically by script and there's no need to manually compare against a reference image [08:18:00.0000] Philip`: ok, two that i can't determinate if it's a pass or a fail, and they are missing reference images (commented as #TODO) [08:19:00.0000] Philip`: and they never complete in the runner [08:19:01.0000] 2d.imageData.put.{clamp,round}? [08:19:02.0000] yes [08:19:03.0000] (I can't see any other TODO-commented ones) [08:20:00.0000] Those really ought to be determined pass/fail by the script [08:20:01.0000] seems they aren't [08:23:00.0000] In any particular browser/version? [08:23:01.0000] I've not seen it fail anywhere before, and don't see why it would start [08:25:00.0000] i'm testing with an opera build [08:25:01.0000] Oh, uh... [08:26:00.0000] Does the page's script have the "# TODO ..." line at the bottom? [08:26:01.0000] like, in JavaScript? [08:26:02.0000] where it's a syntax error [08:26:03.0000] because I had mis-indented my YAML [08:26:04.0000] and I fixed the online HTML files but not uploaded the updated source data [08:27:00.0000] aha [08:27:01.0000] yeah, they have # TODO [08:28:00.0000] If you're building from the YAML source, just remove two spaces before the "# TODO: c" lines and it should be happier [08:28:01.0000] I really should upload a fixed version at some point, but it takes a non-zero amount of effort... [10:04:00.0000] i think should replace
[10:05:00.0000] and just be dropped [10:06:00.0000] (or maybe even drop completely) [10:08:00.0000]
... Ian
is commonly used for that purpose [10:08:01.0000] though, now that's disallowed [10:09:00.0000] Wasn't that
..

Ian

recently ? [10:09:01.0000] yeah, but cite doesn't really give you the desired styling, and makes styling a bit more difficult than having a separate element [10:09:02.0000] krijn: yeah, that also has problems with styling [10:09:03.0000] Yeah [10:12:00.0000] So that's what you get when you try to work with the draft now :) [10:13:00.0000] what? [10:13:01.0000] http://fronteers.nl/blog/2008/01/commissie-diplomering-zoekt-vragen - I tried following the HTML5 examples there [10:13:02.0000] Almost a month ago that was how you should use
and :) [10:14:00.0000] Ow, it still is [10:14:01.0000] Even though the name there isn't [10:14:02.0000] Oh well [10:39:00.0000] http://krijnhoetmer.nl/irc-logs/xhtml/20080220#l-423 [10:42:00.0000] zcorpan: what's "there"? [11:33:00.0000] http://www.tbray.org/ongoing/When/200x/2008/02/19/Odd-Partnerships#c1203530343.647028 [12:25:00.0000] dbaron: did the table implement HTML 4.0 or HTML 4.01? [13:34:00.0000] annevk: see my mail to csswg before you change the reference :-) [13:39:00.0000] thanks [13:39:01.0000] I thought Mark Baker was more pragmatic [13:39:02.0000] :( [13:41:00.0000] and Mozilla dropping credentials in requests is also beyond me [13:44:00.0000] annevk: dropping credentials? [13:45:00.0000] they suddenly want to do what some people have been advocating, not sending cookies or authentication info [13:45:01.0000] no reason has been given [13:46:00.0000] where was this announced? [13:47:00.0000] /me decides not to reply to some e-mails from hsivonen and annevk about
where they're just responding to other e-mails on the subject without really making suggestions for the spec [13:47:01.0000] (e-mails from 2004/2005) [13:47:02.0000] gavin, http://lists.w3.org/Archives/Public/public-appformats/2008Feb/0203.html [13:47:03.0000] Hixie, if I don't raise questions but just give answers there's no need to either quote or reply to me [13:48:00.0000] :-) [13:48:01.0000] yeah, you and henri have both said that before iirc [13:48:02.0000] annevk: http://wiki.mozilla.org/User:Sicking/Cross_Site_XHR_Review provides some rationale [13:49:00.0000] the reasons appear to mostly be "it's safer in case something goes wrong" [13:49:01.0000] it defeats most use cases [13:50:00.0000] I'm not really familiar enough with the usecases to comment [13:50:01.0000] gavin, but yeah, that's not a new reason [13:50:02.0000] gavin, k [13:50:03.0000] you can already send cookies cross-site using
[13:50:04.0000] (I read that page earlier fwiw.) [13:50:05.0000] anything that's going to go wrong is already going wrong [13:51:00.0000] the problem is only reading the data back [13:51:01.0000] under anne's fascist "no fancy headers" scheme, you can't even control the content-type! [13:51:02.0000] :-P [13:51:03.0000] yeah, lets blame me :p [13:51:04.0000] these would be good points to bring up with sicking, I guess [13:51:05.0000] maybe you already have [13:52:00.0000] Hixie, I have since made a new proposal for headers based on suggestions from Henri, but maybe I should do that again in a separate thread [13:52:01.0000] sicking doesn't sound very convinced based on that wiki page [13:52:02.0000] he even himself raises issues with not sending cookies [13:53:00.0000] probably comes from someone else [13:53:01.0000] /me looks at jruderman [13:55:00.0000] last tuesday, sicking and i were the only ones arguing for sending cookies. most of the others in the meeting were arguing for not sending cookies. [13:56:00.0000] wow [13:56:01.0000] it's not very frequent for dan veditz and me to disagree. that was strange. [13:56:02.0000] we spent a whole hour arguing about whether cookies should be sent [13:56:03.0000] if you don't send cookies, it's basicalyl useluss [13:56:04.0000] less [13:56:05.0000] lly [13:57:00.0000] it still works for "public data" without cookies, but yeah [13:57:01.0000] cross-site xhr is not for public data [13:57:02.0000] we can already do that with
...-- Ian