| 03:21 | <G0k> | hgey all |
| 03:23 | <G0k> | is there a way to see the history of changes to an html 5 spec section? |
| 03:56 | <MikeSmith> | G0k - no, not for a specific section, afaik |
| 03:57 | <G0k> | oki, thanks |
| 03:57 | <Dashiva> | You could blame your way through it with svn, but that's a hassle :) |
| 04:11 | <othermaciej> | if you have a specific question, people here might know the answer |
| 04:14 | <G0k> | well in http://www.whatwg.org/specs/web-apps/current-work/multipage/section-event0.html |
| 04:14 | <G0k> | the MessageEvent interface |
| 04:14 | <G0k> | i seem to recall the source being a document |
| 04:14 | <G0k> | now it's a window |
| 04:14 | <G0k> | is this a change or am i misremembering? |
| 04:14 | <othermaciej> | it was changed to work on Window |
| 04:14 | <othermaciej> | discussion is in the whatwg archive |
| 04:15 | <othermaciej> | basically so exceptions to same-origin policy can all be a the Window level |
| 04:16 | <G0k> | so how does that affect like..iframes? |
| 04:17 | <G0k> | if an iframe posts a message to its parent, it sees it as coming from the same window but a different domain? |
| 04:17 | <othermaciej> | iframes have a different Window than their parent |
| 04:17 | <G0k> | oh |
| 04:17 | <G0k> | heh sorry my dom skillz are a little weak |
| 04:19 | <othermaciej> | G0k: this part of the spec explains in detail how Window and Document fit together and work with frames: http://www.whatwg.org/specs/web-apps/current-work/multipage/section-windows.html#windows |
| 10:23 | annevk | still isn't really convinced by the postMessage change |
| 10:23 | <hsivonen> | annevk: do current impls do what you require? |
| 10:25 | <annevk> | Opera has the only implementation |
| 10:26 | <annevk> | Mozilla is implementing and wants changes because then all security exceptions supposedly are on Window |
| 10:26 | <hsivonen> | annevk: oh is this about cross-doc? I thought you were talking about XHR body on GET |
| 10:27 | <annevk> | a) that patch has gone nowhere and b) I'm not convinced there are no exceptions left on Document |
| 10:27 | <annevk> | oh, I'm not sure about that one |
| 10:27 | annevk | hopes to avoid editing the specification |
| 11:17 | <othermaciej> | annevk: in WebKit, it's definitely the nearl all security exceptions are on Window and none are on Document |
| 11:18 | <othermaciej> | (actually I guess it is fair to say that all exceptions are on Window but there are a few security checks in other places) |
| 11:29 | <annevk> | ok |
| 11:29 | <annevk> | I did get claims that .open() and .close() and such had similar things, but maybe that's something else |
| 11:29 | <annevk> | hmm |
| 11:36 | <othermaciej> | window.open() and window.close() are allowed from any other window |
| 11:36 | <othermaciej> | that's not true for most Window functions or properties |
| 11:37 | <othermaciej> | in the case of Document, in WebKit at least you can't even get the document object from a window that's not same-origin |
| 11:37 | <othermaciej> | I think in Gecko you can get it, but it throws on any property access |
| 11:37 | <annevk> | document.open() must have a similar check then |
| 11:37 | <annevk> | as it can be a synonym for window.open() depending on the number of arguments |
| 11:38 | <annevk> | ah, I see |
| 11:38 | <othermaciej> | it can be a synonym but I do not think it is required for web compatibility for document.open() to ever be callable from a window with a different origin |
| 11:38 | <othermaciej> | in WebKit that's not possible because you can't even get to the document object for another window |
| 11:38 | <annevk> | right |
| 11:38 | <annevk> | hmm |
| 11:44 | <hsivonen> | does html5lib have a SAX view to the parse tree? |
| 12:46 | <zcorpan> | hello whatwg |
| 12:46 | <othermaciej> | hello zcorpan |
| 12:46 | <hsivonen> | zcorpan: hello |
| 12:46 | <hsivonen> | zcorpan: do you know what's happening with ARIA datatypes? |
| 12:47 | <hsivonen> | zcorpan: including the use of qnames there |
| 12:47 | <hsivonen> | zcorpan: and the general ugliness of XSD dependencies |
| 12:51 | <zcorpan> | hsivonen: not sure. xhtml2 wg wants curies, i think. pfwg and svgwg seem to want opaque strings, i think |
| 12:53 | <hsivonen> | zcorpan: even for stuff datatype='xsd:anyURI'? |
| 12:54 | <zcorpan> | hsivonen: example of what that would be? |
| 12:55 | <hsivonen> | zcorpan: dunno, but http://www.w3.org/TR/2007/WD-aria-state-20071019/#datatype |
| 12:55 | <hsivonen> | interestingly, the stated datatype is string--not qname |
| 12:56 | <hsivonen> | but the prose says qname |
| 12:56 | <hsivonen> | yay |
| 12:56 | <zcorpan> | "Clearly XSD base types are simpler for the user agent to understand." |
| 12:57 | <annevk> | being a ua rep: no |
| 12:59 | <hsivonen> | I wonder if it is too late to bring in the WF2 datatypes instead |
| 12:59 | <annevk> | where do they use datatypes? |
| 13:00 | <annevk> | ah, I see |
| 13:00 | <othermaciej> | zcorpan: they must have an interesting theoretical model of "the user agent" |
| 13:02 | <hsivonen> | zcorpan: btw, what's the case-(in)sensitivity story for ARIA in text/html? is in going to be lower-case only? |
| 13:03 | <zcorpan> | hsivonen: the values? |
| 13:54 | <hsivonen> | zcorpan: e.g. role='foo' |
| 13:54 | <hsivonen> | zcorpan: is foo case-sensitive? |
| 14:23 | <zcorpan> | hsivonen: i'd say yes |
| 14:23 | <zcorpan> | but i'm not sure if it's defined anywhere |
| 14:24 | <annevk> | i'd say no |
| 14:24 | <zcorpan> | annevk: why? |
| 14:24 | <annevk> | for consistency with other enumerated attributes |
| 14:25 | <zcorpan> | but role isn't really enumerated |
| 14:25 | <zcorpan> | it's more like class="" |
| 14:25 | <annevk> | it has a fixed amount of known values |
| 14:26 | zcorpan | finds a login button in the spec |
| 14:26 | <annevk> | i agree it's slightly different though given that it has fallback |
| 14:26 | <zcorpan> | yeah |
| 14:26 | <zcorpan> | case-sensitive is simpler |
| 14:28 | <annevk> | hmm, the annotating doesn't work in Opera |
| 14:30 | <annevk> | why is event-source considered for removal? |
| 14:31 | <hsivonen> | zcorpan: based on specs, I'd say case-sensitive as well |
| 14:32 | <hsivonen> | annevk: changing the case-senstivity of role from whatever Firefox does is probably counter-productive |
| 14:33 | <hsivonen> | annevk: I haven't tested what it does here |
| 14:34 | <zcorpan> | hmm, does anyone else see the safari icon? e.g. at <canvas> |
| 14:35 | <Philip`> | zcorpan: The login button is for editing annotations |
| 14:35 | <Philip`> | zcorpan: I see that icon in Firefox 2, but not in Opera 9.2 |
| 14:35 | <hsivonen> | hmm. I've reached some kind of zen state. I feel no urge to reply to the <xmp> thread |
| 14:35 | <zcorpan> | Philip`: ok |
| 14:36 | <dglazkov> | hsivonen, aka "apathy"? |
| 14:36 | <annevk> | hsivonen, like selectors, media queries are not "CSS" |
| 14:36 | <hsivonen> | dglazkov: if you want to spin it negatively :-) |
| 14:37 | Philip` | hopes the <xmp> discussion can avoid collapsing into an XHTML discussion |
| 14:37 | <gsnedders> | Philip`: Fail. |
| 14:37 | <Philip`> | It hasn't collapsed that way yet, though it's on the edge :-) |
| 14:37 | dglazkov | doesn't like being negative |
| 14:37 | <dglazkov> | zen it is |
| 14:38 | <gsnedders> | Philip`: Oh, I can see into the future. You can't? |
| 14:43 | <hsivonen> | zcorpan: I see the safari icon |
| 14:44 | <hsivonen> | does the disclosure triangle in spec annotations work in any browser? |
| 14:45 | <Philip`> | hsivonen: Disclosure triangle? I thought the triangle was just pointing at the the attached section (except in Firefox where it gets drawn in the wrong place) |
| 14:45 | <hsivonen> | Philip`: oh. |
| 14:46 | <hsivonen> | Philip`: well, it's a usability bug then |
| 14:46 | gsnedders | still wants to know how he had an account without ever requesting one |
| 14:46 | <hsivonen> | Philip`: because I though it was a broken disclosure triangle |
| 14:46 | <Philip`> | hsivonen: I could be wrong, of course :-) |
| 14:47 | <hsivonen> | Philip`: your explanation makes sense considering the safari rendering |
| 14:47 | <gsnedders> | I always get, "Failed to save data: OK", when trying to save :\ |
| 14:48 | <hsivonen> | gsnedders: the test data is broken in Firefox 2 |
| 14:49 | <hsivonen> | gsnedders: you need a trunk build, it seems |
| 14:49 | <gsnedders> | Doesn't work in Saf3, either |
| 14:50 | <Philip`> | Somebody ought to do something about all this non-interoperability |
| 14:50 | <hsivonen> | gsnedders: in Safari 3 it works at least a bit more than in Firefox 2 |
| 14:50 | <gsnedders> | Philip`: EMCAScript 5? |
| 14:50 | <gsnedders> | DOM5? |
| 14:50 | <annevk> | I would like to know what causes the breakage in Opera |
| 14:50 | <hsivonen> | I wonder if this is the Acid3 test |
| 14:50 | <gsnedders> | does it work in _anything_ except Fx3? |
| 14:51 | <OmegaJunior> | I thought ES4 was in the works.... ES5 doesn't exist yet. |
| 14:51 | <MikeSmith> | annevk - maybe Hallvord Steen could figure out pretty quickly |
| 14:52 | <gsnedders> | OmegaJunior: peh. we have XHTML5 while XHTML2 is in the works! |
| 14:58 | <MikeSmith> | Is something supposed to happen if I click on the triangle in an annotation |
| 15:00 | <Philip`> | MikeSmith: Probably not |
| 15:00 | <MikeSmith> | k |
| 15:00 | <Philip`> | MikeSmith: It seems it's just rendered differently in some browsers so that it no longer makes sense |
| 15:04 | <MikeSmith> | Philip` - ah, yeah, I see now. In Opera and Safari, it renders on the right edge of the annotation box |
| 15:08 | <annevk> | it's displayed on the wrong place in Firefox |
| 15:09 | <annevk> | oh |
| 16:02 | <MikeSmith> | annevk - are you running parallels or vmware on your Mac? |
| 16:05 | <annevk> | neither, I think |
| 16:08 | <MikeSmith> | annevk - I just mention because you said you were using Ubuntu also |
| 16:09 | <MikeSmith> | or some Linux |
| 16:10 | <MikeSmith> | you know if you use Parallels you can run OSX, Linux, Windows on the same machine |
| 16:10 | <MikeSmith> | without needing to reboot |
| 16:11 | <annevk> | you can also use two laptops :) |
| 16:12 | <MikeSmith> | sure, two is better than one of course |
| 16:12 | <MikeSmith> | and three would be even better than two |
| 16:12 | <MikeSmith> | you could hire a valet to help you carry them around |
| 16:13 | <annevk> | actually, I have three |
| 16:13 | <annevk> | although one is from zcorpan |
| 16:14 | <MikeSmith> | you and zcorpan could maybe share a valet |
| 16:14 | <MikeSmith> | alternating days or something |
| 16:18 | <Lachy> | VMWare is better than Parallels. That's what I use on my Mac |
| 16:19 | <gsnedders> | yeah, that's what I concluded too. |
| 16:20 | Philip` | just uses Boot Camp and almost never boots OS X |
| 17:58 | <annevk> | hmm, people use CSS interop bugs as CSS hacks |
| 17:59 | <annevk> | guess there's nothing new there |
| 17:59 | <othermaciej> | isn't that the definition of a CSS hack? |
| 17:59 | <othermaciej> | (except that ideally you should use the failure case as the hack trigger) |
| 18:00 | <annevk> | it's not yet clear here what the hack is and what the failure |
| 18:01 | annevk | was just sending some results of interop testing to www-style |
| 18:01 | annevk | wonders when someone brings accessibility to the <xmp> discussion |
| 18:29 | <annevk> | hmm, I don't agree either with not changing the method names because of some Java API |
| 18:29 | <annevk> | grmbl |
| 18:30 | <othermaciej> | especially since the Java API was made by copying an Editor's Draft, with the agreement that it would not it being a work in progress and subject to change |
| 18:31 | <othermaciej> | now they ask not to change |
| 18:31 | <othermaciej> | that doesn't seem right |
| 18:31 | <annevk> | lengthComputable... |
| 18:57 | <annevk> | ah |
| 18:58 | <annevk> | people are still doing 2DGraphics in Java |
| 18:58 | <annevk> | http://www.cycloloco.com/shadowmaker/ |
| 18:58 | <annevk> | if only shadows were implemented |
| 18:58 | <annevk> | we could <canvas>ify it |
| 19:04 | <Philip`> | <canvas> shadows wouldn't help much when you don't want the shadow to be exactly the same shape as the object |
| 19:05 | <Philip`> | You can do Gaussian blurring with canvas easily, but I can't think of an easy way to convert an image to greyscale |
| 19:06 | <Philip`> | You can't do shears in most implementations either |
| 19:10 | <Philip`> | Oh, actually, maybe greyscale is easy, by drawing a grey rectangle with destination-in |
| 19:12 | <Philip`> | Oh, actually, no current released implementation can do shearing |
| 19:13 | <Philip`> | (unless you fake it with rotations and scales) |
| 19:14 | <Philip`> | Anyway, you should implement canvas shadows anyway ;-) |
| 19:31 | <annevk> | heh |
| 21:18 | <anne-mac> | keeping globalStorage & co makes some sense given that SQL is asynchronous |
| 21:18 | <anne-mac> | I guess that will satisfy most requirements |
| 21:18 | <anne-mac> | btw, othermaciej, did you get around contacting the Firefox guys about changing the design of globalStorage? |
| 21:18 | <anne-mac> | especially its broken security model |
| 21:19 | <anne-mac> | maybe that can be changed in time |
| 21:28 | <othermaciej> | anne-mac: I wasn't planning to contact them, I think Hixie was |
| 21:29 | <anne-mac> | oh |
| 21:32 | <anne-mac> | Hixie, you around? |
| 21:35 | <anne-mac> | interesting, IE8 will have a new rendering engine |
| 21:36 | <anne-mac> | although on the other hand it's sort of problematic if that's going to be opt-in |
| 21:37 | <Lfe> | anne-mac: url? :-) |
| 21:37 | <anne-mac> | depending on what authors code againts, of course |
| 21:37 | <anne-mac> | http://www.molly.com/2007/12/05/conversation-with-bill-gates-about-ie8-and-microsoft-transparency/ |
| 21:38 | <anne-mac> | "MOLLY HOLZSCHLAG: But they’re not letting people talk about it. I do realize that there is a new engine, " |
| 21:38 | <Lachy> | anne-mac, yeah, Chris mentioned that at web directions 2 months ago |
| 21:38 | <anne-mac> | apparently this has been announced earlier at some conference |
| 21:38 | <anne-mac> | right |
| 21:39 | <Lfe> | Well, by new - do they mean new from scratch or just a "version bump"? |
| 21:39 | <gsnedders> | What I think is really interesting is whether it'll replace the current standards mode or not. |
| 21:39 | <othermaciej> | Chris has announced that they are working on a new engine |
| 21:40 | <othermaciej> | where "engine" is pretty strictly restricted to layout, not a full rewrite of everything |
| 21:40 | <gsnedders> | othermaciej: how much of the layout? |
| 21:40 | <Lachy> | I wonder if they've tried to implement the HTML5 parsing algorithm in the new engine |
| 21:41 | <othermaciej> | I dunno |
| 21:41 | <othermaciej> | http://www.sitepoint.com/blogs/2007/10/01/wds07-bonus-feature-chris-wilson-microsoft/ |
| 21:41 | <Lachy> | though given that we're yet to get any significant feedback from MS about the spec, it doesn't seem too likely |
| 21:42 | <anne-mac> | MS has been pretty much "inactive" everywhere apart from CSS |
| 21:42 | <gsnedders> | I guess even the layout itself is where a lot of the major bugs are. |
| 21:42 | <othermaciej> | layout is where one category of major bugs are |
| 21:42 | <anne-mac> | and even there... |
| 21:42 | <anne-mac> | one category :) |
| 21:43 | <gsnedders> | I mean, is the CSS parsing that bad? That probably depends on the border between parsing and layout. |
| 21:43 | gsnedders | doesn't even mention the DOM |
| 21:45 | <othermaciej> | there's still significant CSS parsing bugs, and certainly CSS processing bugs |
| 21:45 | <anne-mac> | yeah |
| 21:46 | <Hixie> | bugs in IE? say it isn't so! |
| 21:46 | <G0k> | i gotta say i fail to understand the "don't break the web" line |
| 21:46 | <othermaciej> | not that Gecko/Presto/WebKit don't have bugs |
| 21:46 | Hixie | is here now btw |
| 21:46 | <G0k> | they're going to have bug fixes which are going to break things, it's ridiculous to expect otherwise |
| 21:47 | <anne-mac> | Hixie, so 1) <event-source> considered for removal?! 2) annotation doesn't work anywhere... 3) ping Firefox about simplying client-side storage? |
| 21:47 | <G0k> | event-source wha? |
| 21:47 | <othermaciej> | G0k is working on implementing event-source for WebKit, so he'd probably be interested in hearing if it is considered for removal |
| 21:48 | <anne-mac> | G0k, don't worry, we'll just set up whatwg2 |
| 21:48 | <Hixie> | 1) it hasn't really had much demand either from authors or from implementors, and a number of issues have been raised suggested we should massively simplify it even if we keep it. it also seems redundant with the two-way network connection stuff and with multipart xhr. but if there really is demand, we could keep it. we just need to prove there is demand, same as any other feature. |
| 21:48 | <Hixie> | 2) elaborate? |
| 21:48 | <Hixie> | 3) yeah i should do that |
| 21:49 | <G0k> | heh well to be honest having both event-source and RemoveEventTarget seemed kinda redundant |
| 21:49 | <othermaciej> | is there a plan for multipart XHR? |
| 21:49 | <anne-mac> | firefox has an impl of that, but I don't really like it |
| 21:49 | <Hixie> | G0k: event-source and removeeventtarget are the same feature |
| 21:50 | <G0k> | heh awz |
| 21:50 | <Hixie> | anne-mac: i've heard people ask for it a lot, at least at google |
| 21:50 | <anne-mac> | Hixie, 2) it doesn't work in Opera, or Safari or stable versions of Firefox |
| 21:50 | <Hixie> | ah |
| 21:50 | <Hixie> | any idea why? |
| 21:51 | <anne-mac> | Hixie, same for event-source at Opera |
| 21:51 | <Hixie> | i don't think i used anything non-standard |
| 21:51 | <anne-mac> | not really |
| 21:51 | <Hixie> | well if there are any bugs in the code do let me know |
| 21:51 | <anne-mac> | debugging it seemed non-trivial |
| 21:51 | <Hixie> | afk bbiab |
| 21:52 | <othermaciej> | Annotation seems to work on in Safari 3 for me |
| 21:53 | <Hixie> | back |
| 21:53 | <othermaciej> | well actually I'm not sure if it is or not, since I don't know what it's supposed to look like |
| 21:53 | <Hixie> | click the login button at the top |
| 21:54 | <anne-mac> | Hixie, it doesn't make much sense to consider <event-source> for removal but not <datagrid> which is in comparison far less successful |
| 21:54 | <G0k> | how is event-source successful? |
| 21:54 | <Hixie> | oh datagrid will be removed too if nobody wants it |
| 21:55 | <Hixie> | but datagrid is much less mature and hasn't been looked at much by implementors yet |
| 21:55 | <Hixie> | and it has no alternative for authors |
| 21:55 | <anne-mac> | G0k, there's one implementation and one partial patch... |
| 21:55 | <G0k> | well two partial patches but i mean has anyone used it for anything? |
| 21:58 | <anne-mac> | editing the annotation in Safari seems to remove it |
| 21:59 | <anne-mac> | just happened to <canvas> |
| 21:59 | <Hixie> | check for js errors |
| 21:59 | <anne-mac> | G0k, there's a public chat client somewhere |
| 22:00 | <anne-mac> | Hixie, sorry, not familiar enough with Safari to get that running |
| 22:00 | <Hixie> | k |
| 22:00 | <G0k> | i guess i do feel like multipart XHR could very well eliminate 90% of the need for event source |
| 22:01 | <G0k> | even though i think the addEventSource API is about 1000x more elegant |
| 22:01 | <gsnedders> | Hixie: Saf shows no JS errors |
| 22:01 | <Hixie> | odd |
| 22:01 | <Hixie> | wonder what's going on |
| 22:01 | <anne-mac> | Opera doesn't do that either and yet it goes wrong with logging in |
| 22:01 | <anne-mac> | it's annoying |
| 22:02 | <Hixie> | what goes wrong? |
| 22:02 | <gsnedders> | anne-mac: you need the debug menu enabled (just google it) then its obvious enough in that menu |
| 22:02 | <anne-mac> | when you log in you get a single . instead of "Logged in as anne." |
| 22:02 | <Hixie> | odd |
| 22:03 | <gsnedders> | anne-mac: what should show that? |
| 22:03 | <Hixie> | that suggests that the login is failing without a message |
| 22:03 | <gsnedders> | anne-mac: login box? |
| 22:04 | <Hixie> | which means x.status != 200 and != 304 |
| 22:04 | <gsnedders> | I get "OK." under the login button… |
| 22:04 | <Hixie> | anne-mac: reload and try again, i've changed the message that should appear |
| 22:04 | <gsnedders> | and trying to save anything results in, "Failed to save data: OK" (this is Saf 3/Mac) |
| 22:05 | <Hixie> | sounds like xhr bugs |
| 22:05 | <gsnedders> | now I get 403 |
| 22:05 | <Hixie> | 403 means you're not logged in |
| 22:05 | <gsnedders> | "Status: Error: OK (403)." |
| 22:05 | <Hixie> | in theory |
| 22:05 | <gsnedders> | that's trying to login. |
| 22:05 | <Hixie> | wait what? |
| 22:05 | <Hixie> | how the hell are you getting 403 OX |
| 22:05 | <Hixie> | OK |
| 22:06 | <Hixie> | that's not one of the things my script sends back |
| 22:06 | <gsnedders> | I try logging in. |
| 22:06 | <gsnedders> | Oh, I get it all right :) |
| 22:06 | <gsnedders> | :P |
| 22:06 | <Hixie> | do you have a login? |
| 22:06 | <gsnedders> | Hixie: it worked in Fx3b1 |
| 22:06 | <gsnedders> | see, this is why I just get other people to write JavaScript for me |
| 22:07 | <Hixie> | what's your username? |
| 22:07 | gsnedders | stares at Philip` |
| 22:07 | <gsnedders> | Hixie: gsnedders |
| 22:07 | <anne-mac> | I get " var node = response.documentElement;" null value around that line |
| 22:07 | <Hixie> | there's no "gsnedders" user registered |
| 22:08 | <gsnedders> | then how on earth did I login in Fx3b1!? |
| 22:08 | <Hixie> | anne-mac: when doing what in what browser? |
| 22:08 | <Hixie> | gsnedders: no idea |
| 22:09 | <Hixie> | there aree 14 users registered, including 3 annes, but you're not one of them :-) |
| 22:09 | <gsnedders> | OK, now I get, "Logged in as gsnedders." |
| 22:09 | <Hixie> | now you have one :-) |
| 22:09 | <anne-mac> | Hixie, when adding an "x" to the demo field in Safari |
| 22:09 | anne-mac | has anne, test, and test1000 registered, he thinks |
| 22:10 | <gsnedders> | Hixie: null value, line 580 of status.js trying to edit |
| 22:10 | anne-mac | got weird messages when registering |
| 22:10 | <Hixie> | the only way that can happen is if you got a 204... hmm... |
| 22:10 | <gsnedders> | and on that note, g'nite |
| 22:10 | <Hixie> | nn |
| 22:11 | <Hixie> | i can't see any way that get-one-annotation can return a 204... |
| 22:11 | <Hixie> | unless... |
| 22:11 | <Hixie> | oh i know what's going on |
| 22:11 | <Hixie> | i bet these browsers are all caching stuff |
| 22:11 | Hixie | adds cache-control: no-cache |
| 22:11 | <Hixie> | try now |
| 22:12 | <Hixie> | (clear your cache first) |
| 22:12 | <Hixie> | though frankly i don't understand what the browsers are caching if this helps it |
| 22:14 | <Hixie> | try now |
| 22:14 | <Hixie> | sorry |
| 22:14 | <Hixie> | had syntax error |
| 22:25 | <roc> | gah |
| 22:25 | <roc> | Colloquy is useless |
| 22:25 | <G0k> | yeah |
| 22:26 | Hixie | waves his irssi banner |
| 22:26 | <G0k> | it's kinda driving me crazy |
| 22:26 | <roc> | oops wrong channel |
| 22:26 | <Dashiva> | irssi isn't cool until it starts using python for scripting, 'cuz xkcd said so |
| 22:26 | <roc> | but thanks for the support |
| 22:31 | <othermaciej> | Colloquy rules! |
| 22:31 | <G0k> | it's pretty and nice except for the fact that it has some really inexplicable bugs |
| 22:32 | <G0k> | like sometimes you join channels and the text is invisible |
| 22:34 | <roc> | right, that's what's biting me |
| 22:38 | <Hixie> | did all the people who were having problems with the annotation script go to have lunch or something? :-) |
| 22:38 | <G0k> | they gave up |
| 22:38 | <anne-mac> | i had some food |
| 22:38 | <anne-mac> | i have issues with the safari inspector thingie not showing up now |
| 22:39 | <Philip`> | They were so enthralled they forgot to return to IRC |
| 22:40 | <Hixie> | anne-mac: we can get it to work in opera if you want instead |
| 22:40 | <Hixie> | let the safari guys worry about safari :-) |
| 22:40 | Philip` | wonders who will be the first to dare complain it doesn't work in IE |
| 22:41 | <anne-mac> | Hixie, when logging in I get "Error (0)" |
| 22:41 | <anne-mac> | in Opera |
| 22:41 | <Hixie> | in the login dialog? |
| 22:42 | <Hixie> | in an alert? |
| 22:42 | <anne-mac> | in the dialog |
| 22:42 | <Hixie> | and does it say exactly "Error (0)" or does it say "Error: (0)"? |
| 22:42 | <anne-mac> | the latter |
| 22:42 | <anne-mac> | followed by a . |
| 22:42 | <Hixie> | well then that's an opera bug. that means xhr isn't returning a sensible status code. |
| 22:42 | <Hixie> | the 0 is x.status at readystate = 4 |
| 22:43 | <anne-mac> | weird |
| 22:43 | <anne-mac> | thanks though |
| 22:45 | <Hixie> | (does the xhr spec ever allow 0 in rs4? |
| 22:45 | <Hixie> | ) |
| 22:45 | <anne-mac> | if http does |
| 22:45 | <anne-mac> | actually, it does |
| 22:45 | <anne-mac> | or maybe not |
| 22:46 | <G0k> | hixie: so has anything been written w/r/t multipart XHR? |
| 22:46 | <Hixie> | not to my knowledge |
| 22:46 | <anne-mac> | no, not |
| 22:46 | <Hixie> | i'm trying to let anne deal with xhr issues |
| 22:46 | <Hixie> | as i have plenty on my plate already :-) |
| 22:46 | <G0k> | anne, let's talk xhr |
| 22:46 | <anne-mac> | sure |
| 22:47 | <Hixie> | anne-mac: if it turns out to be a bug in my script, please let me know. but i have no idea what it couldf be. |
| 22:48 | <anne-mac> | Hixie, does your script happen to use redirects on the server? |
| 22:49 | <G0k> | so it seems me like what we really want is more control over the the downloading process |
| 22:50 | <anne-mac> | http://dev.w3.org/2006/webapi/XMLHttpRequest-2/ has most of the new stuff, but I'm still awaiting conforming implementations of XHR1 |
| 22:51 | <anne-mac> | G0k, more control, how? |
| 22:52 | <G0k> | like preventing connections from timing out, getting around the 2 connection limit, checking up on the connection download progress |
| 22:52 | <Hixie> | anne-mac: no, no redirects as far as i know |
| 22:52 | <Hixie> | (except maybe in the multipage version?) |
| 22:52 | <Hixie> | afk, brb |
| 22:57 | <anne-mac> | G0k, I'm not sure how to do one and two, but three is addressed |
| 22:57 | <G0k> | uhm. how about just allowing authors to set the connection timeout? |
| 22:58 | <anne-mac> | ah yeah, that's been suggested |
| 22:58 | <anne-mac> | send(data, timeout) or something like that |
| 22:59 | <G0k> | that sounds like progress yes. |
| 22:59 | <anne-mac> | I've only seen one or two requests for that so far |
| 23:00 | <anne-mac> | and it's doable using setTimeout ... (although maybe slightly more annoying) |
| 23:01 | <G0k> | yeah but that sucks |
| 23:01 | <G0k> | you basically end up polling the thing |
| 23:06 | <G0k> | plus...i mean i guess it's an implementation issue but abort() isn't very reliable |
| 23:06 | <anne-mac> | in general we don't want to introduce new features just because others are badly implemented |
| 23:08 | <G0k> | yeah fair enough |
| 23:09 | <G0k> | i mean i feel like, in principle, we should be able to implement dom events with XHR |
| 23:09 | <G0k> | should be a design goal, that |
| 23:09 | <anne-mac> | dom events? |
| 23:09 | <G0k> | er...server-sent dom events |
| 23:09 | <G0k> | if they're really going bye bye |
| 23:10 | <anne-mac> | i also think that in general overloading xhr is a bad idea |
| 23:10 | <anne-mac> | it's already pretty bloated thanks to historic accidents |
| 23:10 | <G0k> | i also kinda feel like the name is sorta crappy |
| 23:10 | <G0k> | why not just make it HttpRequest? |
| 23:10 | <anne-mac> | with weird events being dispatched at random places |
| 23:11 | <anne-mac> | that would be similar to renaming <html> to <web> |
| 23:11 | <G0k> | but....no one uses it for xml |
| 23:12 | <anne-mac> | i'll try again; it already works, why break that? |
| 23:13 | <G0k> | just because it works doesn't mean that a new API wouldn't be clearer, less buggy, and more featureful |
| 23:13 | <anne-mac> | sounds like xhtml2 to me |
| 23:13 | <Hixie> | a new api means two apis |
| 23:13 | <Hixie> | two apis, one of which sucks, is worse than one api which sucks. |
| 23:14 | <G0k> | well there's already about 6 APIs for drawing a box |
| 23:14 | <Hixie> | it means more spec work, more implementation work, more testing, more tutorials, etc |
| 23:14 | <anne-mac> | G0k, that should not be an excuse to make it worse! :) |
| 23:14 | <Hixie> | bad things exist. they aren't a license to add more bad things. |
| 23:15 | <Hixie> | (the same arguments suggest removing event-source and co, btw) |
| 23:15 | <G0k> | so i presume <canvas> is going away too? :) |
| 23:15 | <anne-mac> | (i'm not convinced) |
| 23:17 | <Hixie> | anne-mac: that's why it's not already gone :-) |
| 23:17 | <G0k> | i mean we have to break backwards compatibility anyway to implement certain features |
| 23:18 | <Hixie> | G0k: how do you create, say, a fractal viewer, without <canvas>? |
| 23:18 | <G0k> | extreme SVG use? |
| 23:19 | <Hixie> | uh huh |
| 23:19 | <Philip`> | Lots and lots of tiny <div>s |
| 23:19 | <Hixie> | without taking down the browser. |
| 23:19 | <Hixie> | and without abusing semantics. |
| 23:19 | <Philip`> | That's just a quality-of-implementation issue |
| 23:19 | <Philip`> | <div> has no semantics :-) |
| 23:20 | <inimino> | Hixie: http://claudiusmaximus.goto10.org/GG/graphgrow.svg |
| 23:20 | <G0k> | and if we're going into semantics, then continuing to use the XMLHttpRequest name is especially ridiculous |
| 23:21 | <Hixie> | semantics are related to definitions, not names |
| 23:21 | <inimino> | (I think it would be more better with <canvas> though) |
| 23:22 | anne-mac | wonders why that doesn't show anything in Opera |
| 23:23 | inimino | looks for the link to the description page... |
| 23:23 | <G0k> | even then, the XHR definition of "scripted client functionality for transferring data between a client and a server" is not quite right |
| 23:25 | <G0k> | in any case, retaining backwards compatibility with oddities in the old ActiveX control's API seems equivalent to adding the marquee and blink tags back in |
| 23:27 | <G0k> | anyway i'm off for a bit |
| 23:29 | <inimino> | http://claudiusmaximus.blurty.com/#419 |
| 23:30 | <inimino> | anne-mac: looks like it was tested only in Fx |
| 23:31 | <inimino> | when it works you can build fractals like this http://mjclement.com/images/screenshots/2007-12-09_4.png |
| 23:31 | <inimino> | and it uses SVG + JavaScript |
| 23:31 | <anne-mac> | interesting |
| 23:32 | <inimino> | I think an implementation using <canvas> could be faster |
| 23:32 | <anne-mac> | i wonder how the bugs are distributed |
| 23:32 | <inimino> | I didn't look at the code but I assume the SVG ends up rather large |
| 23:36 | <anne-mac> | there's some script error in Opera |
| 23:36 | <anne-mac> | looks like something is null/undefined where an object is expected |
| 23:36 | <anne-mac> | oh well, too late |
