| 01:59 | <Lachy> | Firefox seems to have some strange behaviour with document.write in this case: |
| 02:00 | <Lachy> | <script>document.write("<\meta charset=UTF-8>");</script> doesn't detect has UTF-8, but |
| 02:00 | <Lachy> | <script>document.write("<\meta charset=UTF-8");</script>> does. |
| 02:01 | <Hixie> | good times |
| 02:02 | <Lachy> | AFAICT, according to the spec, both should be treated as UTF-8, since document.write just injects directly into the input stream |
| 02:02 | <Hixie> | yup |
| 02:03 | <Philip`> | What about <script>unknownfunction("<\meta charset=UTF-8>");</script> ? |
| 02:03 | <Philip`> | Oh, wait, I read you the wrong way around |
| 02:04 | <Philip`> | Sounds like it only does the charset processing if the meta element is created through the normal non-document.write parser, or something? |
| 02:05 | <Lachy> | this works too: <script>document.write("<");</script>meta charset=UTF-8> |
| 02:06 | <Lachy> | looks like you can split it anywhere inside the string, and it only fails if the whole element is written with .write() |
| 02:08 | <Lachy> | my next challenge is to detect the current encoding the browser is using, using document.write() to output a conflicting encoding, trigger a reparse, and then detect and output a different one :-) |
| 02:34 | <Hixie> | i added relative checkin activity to the issues graph |
| 08:09 | <Hixie> | man, coming up with a good description of what a <label> element represents is non-trivial |
| 08:44 | <annevk> | <label> represents a form control label |
| 08:52 | <Hixie> | So what is the form control in <label>Foo</label> ? |
| 08:53 | <annevk> | there's none because it's not associated with any? |
| 08:54 | <Hixie> | well then it's not a form control label |
| 08:54 | <annevk> | <label> represents a label that is to be associated with a form control |
| 08:55 | <annevk> | using either an explicit reference (for="") or indirect (nesting |
| 08:55 | <annevk> | ) |
| 08:55 | <Hixie> | there you're using the name of the element to define the element, which is poor form since it reads badly (repetition) and doesn't help the author who didn't know what "label" meant in the first place :-) |
| 08:58 | <hsivonen> | Hixie: but the string "label" in the element name is an opaque string and the word "label" in the defining prose has a meaning in English |
| 08:58 | <annevk> | hmm, those authors can look at the examples |
| 08:58 | <hsivonen> | Hixie: surely it is ok to use the word "title" to define <title> |
| 08:58 | <Hixie> | you'll note that i also use another name |
| 08:58 | <Hixie> | er, another word |
| 08:58 | <Hixie> | namely "name" |
| 08:58 | <Hixie> | to define <title> |
| 08:58 | <Hixie> | similarly <caption> is defined using the word "title" |
| 08:59 | <Hixie> | and <label> using the word "caption"... |
| 08:59 | <annevk> | maybe that will confuse authors |
| 08:59 | <Hixie> | i doubt it. i posit that many authors who are looking up an element definition are looking it up because the name of the element wasn't clear to them, so i try to avoid reusing the name of the element in its definition. |
| 08:59 | <annevk> | http://search.twitter.com/search?q=whatwg :) |
| 09:00 | <hsivonen> | Hixie: if guessing from the name is ok, expect problems with <dialog> and <details> |
| 09:00 | <Hixie> | whether it's ok or not, people will do it |
| 09:01 | <Hixie> | thankfully for <dialog> and <details>, the problem is mitigated |
| 09:01 | <Hixie> | for <dialog> because of the content model, so that authors will usually come across it in the right usage |
| 09:01 | <Hixie> | and for <details> because if its default rendering |
| 09:04 | <Hixie> | ah well |
| 09:04 | <Hixie> | <input> is next |
| 09:04 | <Hixie> | better go to bed! |
| 09:04 | <Hixie> | nn |
| 09:04 | <annevk> | :) nn |
| 10:08 | <hsivonen> | it annoys me that tweets with tiny urls don't show where the URLs lead |
| 10:09 | <annevk> | search.twitter has an (expand) function |
| 10:13 | <hsivonen> | can I subscribe to an expanded version of any twitter feed? |
| 10:14 | <annevk> | dunno |
| 10:15 | <jgraham> | It would be nice if twitter allowed hyperlinks and didn't count the characters in the tag |
| 10:16 | <jgraham> | like <a href="http://example.com/my/long/url/">name</a>; being 4 characters |
| 10:17 | <Dashiva> | I'm not sure SMS operators would allow you free bytes like that ;) |
| 10:17 | <MikeSmith> | on the plus side, twitter has helped to create a better market for URL-shortening services |
| 10:18 | <Philip`> | How about (revolutionary idea) not applying silly SMS limits to web-based entry? |
| 10:18 | <jgraham> | Hmm I guess SMS would be a problem |
| 10:18 | <Philip`> | (Does anyone even use Twitter over SMS?) |
| 10:18 | <annevk> | I believe so |
| 10:18 | <annevk> | not me though |
| 10:18 | <MikeSmith> | twitter over SMS doesn't work any more |
| 10:18 | <annevk> | oh |
| 10:18 | <MikeSmith> | they shut it down temporarily |
| 10:18 | <MikeSmith> | many months ago |
| 10:18 | <jgraham> | MikeSmith: They still all obscufate the destination though |
| 10:18 | <Lachy> | MikeSmith, in all countries? |
| 10:18 | <MikeSmith> | temporarily |
| 10:18 | <jgraham> | by necessity |
| 10:19 | <jgraham> | (the URL shortening services) |
| 10:19 | <Lachy> | I thought twitter SMS was still working in Aus |
| 10:20 | <MikeSmith> | Lachy: don't work for me at least.. but I don't actually use SMS, I use the XMPP thing |
| 10:20 | <Lachy> | http://blog.twitter.com/2008/08/changes-for-some-sms-usersgood-and-bad.html |
| 10:20 | <MikeSmith> | but I thought they use they same backend for both |
| 10:20 | <Lachy> | says they stopped in some European places |
| 10:21 | <annevk> | anyway, placing a limit prevents people from writing essays |
| 10:21 | <Lachy> | I wish they would restore Jabber support, that's been gone for ages and I need it |
| 10:21 | <annevk> | counting < as four characters and such however is something they should probably stop with now |
| 10:22 | jgraham | is still amazed by the use of the word need in relation to twitter |
| 10:22 | <MikeSmith> | Lachy: yeah, the Jabber interface is what I had been using, over a bitlbee gateway so that it just shows up in my IRC client |
| 10:23 | <Philip`> | Twitter wouldn't be Twitter if it didn't limit you to very short messages, but limiting URL length too is just a technical limitation that makes things awkward for readers and writers, rather than a useful social limitation that defines the service |
| 10:23 | <Lachy> | MikeSmith, does it still work for you? |
| 10:23 | <MikeSmith> | Lachy: nope |
| 10:23 | <Lachy> | what do you use now? |
| 10:24 | <MikeSmith> | for posting, I use Vim with a pipe to a python script that posts to both twitter and identi.ca |
| 10:24 | <MikeSmith> | for reading, I use the web |
| 10:24 | <Lachy> | I think the 140 character limit for twitter is the most annoying thing about twitter |
| 10:25 | <Lachy> | I understand why they have the limit for SMS reasons, but even for other clients, they restrict the output, which is totally unnecessary |
| 10:25 | <MikeSmith> | I agree with Philip` and annevk about the limit. Yeah, it's arbitrary but having some limit it fundamental to the whole idea of the service |
| 10:26 | <MikeSmith> | but they really should have a better way of handling URLs |
| 10:27 | <Lachy> | yeah, but 140 characters isn't enough. They should double it, and the whole SMS system needs to be upgraded to remove the 160 char limit |
| 10:27 | <zcorpan_> | they could use their own url-shortener when sending sms |
| 10:28 | <MikeSmith> | well, SMS sucks massively anyway and should just be abandoned and replaced by SMTP/MIME e-mail |
| 10:28 | <MikeSmith> | SMS is a technology that never should have existed to begin with |
| 10:28 | <Philip`> | Maybe the danger is that if they started allowing arbitrary-length URLs, they couldn't claim technical reasons for the 140-character limit, and everyone would moan about how restrictive it is, and the Twitter people couldn't say "that's because you don't understand what's good for you, and if we let you write much longer posts then readers would need a longer attention span and the service would collapse" |
| 10:29 | <zcorpan_> | or people would write their messages as URLs |
| 10:29 | <MikeSmith> | heh |
| 10:29 | <Dashiva> | Philip`: Alas, you might be onto something |
| 10:30 | <Dashiva> | I started becoming suspicious when I saw they say they count characters and not bytes |
| 10:30 | <Philip`> | zcorpan_: Would that be a problem if they didn't display the URL, only a short text link, and counted the length of the link text (not the URL) towards the 140 character limit? |
| 10:30 | <Lachy> | MikeSmith, no, SMS is a good thing. MMS, however, is stupid and should be abandoned everywhere |
| 10:30 | <Philip`> | You could still mouse-over the link to read the URL in your browser's status bar, but that's a bit rubbish |
| 10:30 | <Philip`> | so I assume people wouldn't do tht |
| 10:30 | <Philip`> | s//a/ |
| 10:53 | <hsivonen> | Lachy: MMS is like circuit-switched 3G video calls: people figure out pretty quickly that the same device does the "same" thing in the best-effort packet-switched mode much more cheaply and well enough |
| 10:54 | hsivonen | tries to resist starting his broken record about Skype on EDGE/HSDPA |
| 10:56 | <hsivonen> | btw, how's the geolocation stuff going? when should I expect Fennec on N800+GPS to work as a real-time mapping client to Google Maps? |
| 11:21 | <othermaciej> | wow I didn't notice that @headers thing when it first came up |
| 11:22 | <annevk> | hsivonen, there's a draft for geolocation but no Working Group to publish it |
| 11:22 | <othermaciej> | in case anyone is curious, I can state for the record that the Decision Policy was specifically designed with the intent of preventing technical decisions from being made in telecons |
| 11:22 | <othermaciej> | can't tell from latest email if the supposed Chair decision has been withdrawn or not, but it had better be |
| 11:23 | <annevk> | othermaciej, I believe some kind of further clarification was forthcoming after more study from the chairs |
| 11:24 | <othermaciej> | I hope the clarification includes "we're sorry for not following the Decision Policy in the charter and will do so in the future" |
| 11:25 | zcorpan_ | wonders how <script defer> is specced in html5... http://www.stevesouders.com/blog/2008/09/11/delayed-script-execution-in-opera/ |
| 11:25 | <othermaciej> | I believe HTML5 requires you to wait for the document to be loaded on <script defer>, but also includes <script async> |
| 11:26 | <othermaciej> | http://www.whatwg.org/specs/web-apps/current-work/#script |
| 11:26 | <zcorpan_> | how does document.write interact with those? |
| 11:28 | <othermaciej> | you mean if you document.write from a script loaded using one of those mechanisms? |
| 11:28 | <othermaciej> | or if you document.write such a script? |
| 11:28 | <othermaciej> | or if you document.write before or after starting the load, or what? |
| 11:28 | <Philip`> | "A feature of Opera’s “Delayed Script Execution” option is that, even though scripts are deferred, document.write still works correctly. Opera remembers the script’s location in the page and inserts the document.write output appropriately." - is that really true, or does it just reparse the document (and the initial scriptless parse is just to prefetch the scripts and images without blocking)? |
| 11:28 | <zcorpan_> | if you document.write from a script loaded using one of those mechanisms |
| 11:29 | zcorpan_ | doesn't know exactly how dse works |
| 11:30 | <Philip`> | (I'm fairly sure I've heard that IE8 does speculative parsing past synchronous scripts, to prefetch the later included scripts) |
| 11:30 | <Philip`> | (but I have no idea which other browsers do similar things) |
| 11:30 | <zcorpan_> | like webkit |
| 11:31 | <othermaciej> | webkit does speculative preload |
| 11:31 | <othermaciej> | but we won't parse or display while blocked on actually running a script |
| 11:31 | <othermaciej> | doing so would cause bad results unless you reparse after running the script, I expect (which means you couldn't run any further scripts in the speculative rendering, even if inline) |
| 11:33 | <othermaciej> | anyway I think per html5 spec, <script async> document.write would write to a random place, and <script defer> document.write would clobber the document |
| 11:33 | <othermaciej> | ah, I see, that page even explains Delayed Script Execution |
| 11:34 | <othermaciej> | WebKit has sort of the equivalent of that feature too, "low-bandiwidth rendering" |
| 11:34 | <othermaciej> | it does a quick render w/o waiting for scripts or styles |
| 11:34 | <othermaciej> | and then reparses for real |
| 11:34 | <othermaciej> | but that's not even compiled in by default |
| 11:38 | <annevk> | document.write() would just do what it normally does |
| 11:38 | <annevk> | in html5 |
| 11:38 | <othermaciej> | what do you mean "what it normally does" |
| 11:39 | <othermaciej> | <script async> is a newly invented feature so at least in that case there is no pre-existing "what it normally does" |
| 11:39 | <annevk> | it would be the same as executing a script at the point <script async> would be executed |
| 11:40 | <annevk> | (that is, document.write() does not suddenly throw or something because it runs from <script async> or <script defer>) |
| 11:40 | <othermaciej> | yeah, but from <script async> I think it would insert characters at a completely unpredictable place in the document |
| 11:41 | <othermaciej> | and thus should never be used from <script async> |
| 11:41 | <annevk> | oh, I agree |
| 11:41 | <othermaciej> | I think what (some) people want is for document.write to magically write after the point where the <script> tag occurs |
| 11:41 | <othermaciej> | not realizing that you can't consistently both do that and not block the parser, short of reparsing later |
| 11:42 | <annevk> | for the <script src> case you can do some magic |
| 11:43 | <othermaciej> | seems to me there is no way to get guaranteed identical results without reparsing the rest of the document after the script |
| 11:44 | <othermaciej> | (which is what it sounds like the Opera feature does, though maybe it optimizes some cases where it can prove reparsing is not necessary; I wouldn't know) |
| 11:44 | <zcorpan_> | Hixie: http://www.w3.org/Bugs/Public/show_bug.cgi?id=6057 is a dup of http://www.w3.org/Bugs/Public/show_bug.cgi?id=5979 |
| 11:51 | <zcorpan_> | http://www.w3.org/mid/830059.6331.qm⊙wmsyc -- hmm, perhaps someone should point him to whatwg⊙wo |
| 12:00 | <MikeSmith> | hsivonen: you have problems with Skype on HSDPA? |
| 12:05 | <zcorpan_> | Philip`: perhaps the spec splitter should put the forms stuff on its own page? |
| 12:08 | <hsivonen> | MikeSmith: the problem I have is that I don't have an S60 app |
| 12:09 | <hsivonen> | MikeSmith: but my point is that Skype from N800 works across the packet data feature of my phone |
| 12:09 | <hsivonen> | MikeSmith: so it's kinda pointless to try to sell the features as circuit switched to me |
| 12:09 | <MikeSmith> | ah |
| 12:10 | <hsivonen> | circuit switched is overrated |
| 12:14 | <MikeSmith> | most innovations in telecom technologies are overrated |
| 12:15 | <MikeSmith> | some are literally overrated |
| 12:16 | <hsivonen> | I've had a video call enabled phone for a year, and I haven't used it for video call even a single time |
| 12:18 | <MikeSmith> | I think the main use that video calls on mobiles is being put to is for interactive porn services |
| 12:18 | <zcorpan_> | Hixie: <body background> is supposed to be defined to contain an URL as far as reflecting attributes go, right? |
| 12:18 | <zcorpan_> | s/an/a/ |
| 12:51 | <Philip`> | zcorpan_: That sounds sensible, and should be easy to do |
| 12:52 | <Philip`> | (but I'm busy with other things at the moment, so I won't fix it now) |
| 13:04 | <MikeSmith> | http://www.stevesouders.com/blog/2008/09/11/delayed-script-execution-in-opera/ |
| 13:05 | <MikeSmith> | "If browsers supported an implementation of SCRIPT DEFER that behaved similar to Opera’s “Delayed Script Execution” feature, we’d all be better off." |
| 13:05 | <MikeSmith> | othermaciej: doesn't Webkit already have something like delayed script execution? |
| 13:06 | <MikeSmith> | I remember the Android announcement mentioning something that they had added |
| 13:06 | MikeSmith | finds WTF_USE_LOW_BANDWIDTH_DISPLAY |
| 13:12 | MikeSmith | finds https://lists.webkit.org/pipermail/webkit-dev/2008-February/003341.html |
| 13:13 | <zcorpan_> | MikeSmith: check the logs from 2h back :) |
| 13:13 | <MikeSmith> | "-DWTF_USE_LOW_BANDWIDTH_DISPLAY=1 is not so good. The loader in current builds is significantly faster and better at scheduling than those in other mobile browsers so I don't think we need hacks like this." |
| 13:13 | <MikeSmith> | zcorpan_: OK |
| 13:19 | <Lachy> | I disagree with the conclusion of that delayed script execution article. |
| 13:19 | <Lachy> | If people would stop using document.write at all, we'd all be better off |
| 13:20 | <Lachy> | besides, innerHTML provides all the benefits of using string manipuation to populate the DOM, without the constraints of document.write |
| 13:21 | <virtuelv> | Lachy: expecting that is like pissing upwind |
| 13:21 | <virtuelv> | futile, and likely to cause a mess |
| 13:27 | <Lachy> | virtuelv, I'm not saying we should drop support for it, but encourage authors that want to improve script loading times to avoid it |
| 13:52 | <zcorpan_> | hmm. should DOMParser and XMLSerializer be specced in dom core? |
| 13:54 | <annevk> | might make sense |
| 14:14 | <othermaciej> | MikeSmith: yes, I believe I mentioned the low bandiwdth display, though I got the name wrong |
| 14:14 | <MikeSmith> | othermaciej: yeah, I just re-read through the channel logs on the webkit-dev e-mail thread |
| 14:15 | <othermaciej> | MikeSmith: I think with our speculative pre-parsing and speed in general it is probably not so good even for mobile, but you can certainly get long delays before seeing anything on mobile WebKit browsers/devices that don't use it (like iPhone) |
| 14:17 | <MikeSmith> | othermaciej: what Alp says about displaying a throbber sounds like something worth considering |
| 14:18 | <MikeSmith> | [[ |
| 14:18 | <othermaciej> | MobileSafari UI ideas are something I am not really free to discuss publicly but I'll take that comment under advisement |
| 14:18 | <MikeSmith> | I actually think a throbber or other load notification system is much |
| 14:18 | <MikeSmith> | more effective. When users see unstyled content (HTML without CSS), they |
| 14:18 | <MikeSmith> | immediately consider it a bug in the browser. Not good. |
| 14:18 | <MikeSmith> | ]] |
| 14:18 | <MikeSmith> | understood |
| 14:35 | <zcorpan_> | <input type=password name=pw> <input type=password confirm=pw name=cpw> |
| 14:36 | <hsivonen> | http://broadcast.oreilly.com/2008/09/cross-platform-apis-to-be-in-t.html |
| 14:38 | <othermaciej> | hsivonen: I love when people who almost understand how technology works try to give advice to those actually creating the technology |
| 14:38 | <hsivonen> | does "WTF" in the WebKit define carry its usual meaning or is WTF a special technical term in WebKit? |
| 14:39 | <othermaciej> | WTF stands for Web Template Framework |
| 14:39 | <othermaciej> | has a sort of similar layer in our stack to NSPR for Mozilla |
| 14:39 | <othermaciej> | *similar role |
| 14:40 | <hsivonen> | ok |
| 14:40 | <othermaciej> | but among other things it is where platform conditional macros get defined |
| 14:40 | <othermaciej> | (however, if you read some of the code you may choose to take WTF in a somewhat different sense, since it does make very liberal use of C++ templates...) |
| 14:40 | <wilhelm> | That would be Widget Test Framework at the Opera HQ. (c; |
| 14:40 | Philip` | assumes the choice of a term that acronymises to "WTF" was not purely accidental |
| 14:42 | <virtuelv> | probably not |
| 14:45 | <othermaciej> | I actually wanted to call it WebKit Template Library |
| 14:45 | <othermaciej> | but WTL was taken, by Microsoft |
| 15:02 | <hsivonen> | so if CodeWeavers can get Chromium run on OS X with Wine, they have to have somme solution for multiple processes sharing a window |
| 15:03 | <Philip`> | Does it run multiple OS X processes, or does Wine emulate Win32 processes within a single host process? |
| 15:12 | <hsivonen> | at least the wine initialization takes a long, long time |
| 15:12 | <hsivonen> | I wonder what it does during all that time |
| 15:15 | <hsivonen> | Philip`: there are multiple processes called 'wineloader' |
| 15:15 | <Philip`> | Perhaps it connects with rdesktop to an online Windows server that is executing Chrome |
| 15:15 | <hsivonen> | one process called CrossOver Chromium |
| 15:15 | <hsivonen> | and one wineserver |
| 15:24 | <hsivonen> | perhaps the CrossOver Chromium process runs an X server and the wineloader processes are X11 clients |
| 15:55 | <annevk> | jgraham, so did you manage to get lxml running on DreamHost? |
| 16:12 | gsnedders | wonders whether he can regen HTML 5 without dupe dfn errors after all the work |
| 16:12 | <gsnedders> | Answer: yes. |
| 16:20 | <kangax> | Does anyone know of a publicly available performance test suite for canvas? |
| 16:28 | <Philip`> | kangax: No |
| 16:29 | Philip` | would be interested in one, but isn't aware of anything testing more than a couple of functions |
| 16:36 | <kangax> | Philip`: I see |
| 16:49 | <annevk> | Philip`, when you installed Python on philip.html5.org did it take a long time? |
| 16:51 | <Philip`> | annevk: I don't believe I've ever installed Python there |
| 16:51 | <annevk> | ah ok |
| 16:51 | annevk | is trying it out now |
| 16:52 | annevk | wants to build this Anolis Web service |
| 16:55 | <hsivonen> | hmm. the downside of gandi.net being French is that France doesn't have the most Internet business friendly laws around |
| 16:57 | <Philip`> | The advantage is that if you do something illegal and they want to arrest you, it's fine as long you don't go to France |
| 16:58 | <hsivonen> | France is one of the places I like to go |
| 16:58 | <gsnedders> | Philip`: Ever heard of extradition? |
| 17:01 | <annevk> | yay, python 2.5.1 installed |
| 17:02 | <annevk> | gsnedders, for lxml I also libxml and such right? |
| 17:03 | <gsnedders> | annevk: yeah |
| 17:05 | <hsivonen> | I think I might try putting html5.validator.nu in France while keeping most of my sites in Finland |
| 17:06 | <hsivonen> | but seriously, this virtualized data center thing needs to have better protections for tenant to make counterclaims of legality instead of someone just flipping off a switch |
| 17:07 | <hsivonen> | well, I suppose those better protections cost more money |
| 17:10 | <annevk> | christ, which version of libxml2 to pick? |
| 17:11 | gsnedders | just used the version with his computer :P |
| 17:12 | <gsnedders> | annevk: How about the latest of both libxml2 and libxslt |
| 17:13 | <annevk> | yes, but platform etc. |
| 17:14 | <annevk> | anyway, I found some easy to follow guidelines |
| 17:16 | <annevk> | ok, so libxml and libxslt seem to be working out, great |
| 17:18 | <annevk> | heh, html5.org uses twice as much bandwidth as annevankesteren.nl |
| 17:27 | <annevk> | gettin this "xml toolchain" to run takes ages |
| 17:27 | <annevk> | s/gettin/getting/ |
| 17:29 | <Hixie> | document.write() with <script async> doesn't insert data in the stream at random |
| 17:29 | <Hixie> | i went to extreme lengths to make it always work deterministically |
| 17:57 | <Philip`> | annevk: If you had sensible (root) access and a sensible Linux distribution then lxml would be trivial to install :-) |
| 18:03 | <annevk> | well, easy_install works now, but now I get a bunch of errors like this "src/lxml/lxml.etree.c: In function `__pyx_pf_4lxml_5etree_14_ParserContext___dealloc__'" |
| 18:03 | <annevk> | when trying to easy_install lxml |
| 18:03 | <annevk> | and things like "src/lxml/lxml.etree.c:102485: error: dereferencing pointer to incomplete type" |
| 18:04 | <annevk> | import libxml2 and import libxslt both work |
| 18:05 | <Philip`> | annevk: The "In function ..." line is not the error message - it's just before or after or something |
| 18:05 | <Philip`> | Probably the first actual error is the important one, and the rest might just be consequences of that |
| 18:06 | <annevk> | k, starts with "src/lxml/lxml.etree.c:150:31: libxml/schematron.h: No such file or directory" |
| 18:06 | <Philip`> | Ah, that sounds like the problem |
| 18:06 | <Philip`> | Where did you install libxml to? |
| 18:07 | <annevk> | to $HOME/opt |
| 18:07 | <annevk> | Python is in there too |
| 18:08 | <annevk> | I have this copy of libxml2: ftp://ftp.gnome.org/pub/GNOME/sources/libxml2/2.6/libxml2-2.6.16.tar.gz |
| 18:08 | <Philip`> | You probably need to somehow tell easy_install (or whatever is doing the compiling of lxml) about that location, else it won't find the installed libraries |
| 18:12 | <annevk> | that's not the problem |
| 18:12 | <annevk> | my copy of libxml does not have schematron.h |
| 18:12 | <Philip`> | Oh, okay |
| 18:13 | <annevk> | at least, in opt/include/libxml2/libxml/ I can't find it |
| 18:14 | <annevk> | sigh |
| 18:22 | <gsnedders> | annevk: Did you install libxslt? |
| 18:23 | <annevk> | yes |
| 18:23 | <annevk> | this version of libxslt: ftp://ftp.gnome.org/pub/GNOME/sources/libxslt/1.1/libxslt-1.1.12.tar.gz |
| 18:23 | <annevk> | maybe both are too old? |
| 18:24 | annevk | copied the URIs from somewhere |
| 18:25 | <Philip`> | ftp://ftp.gnome.org/pub/GNOME/sources/libxml2/2.6/libxml2-2.6.30.tar.bz2 has schematron.h |
| 18:26 | <annevk> | thanks, I was just trying to download a newer version |
| 18:26 | Philip` | isn't sure why that's the latest version there, when he has 2.6.32 installed |
| 18:27 | <Philip`> | 2.6.16 is four years old |
| 18:27 | <Philip`> | so it seems reasonable for that to not work |
| 18:31 | <annevk> | still, it's a minor point release |
| 18:31 | <annevk> | oh well, lesson learned |
| 18:31 | <gsnedders> | Heck, 2.7.1 is katest |
| 18:31 | <Philip`> | I guess the policy might be that minor point releases can add API, they just can't remove or incompatibly change existing API |
| 18:32 | <gsnedders> | *latest |
| 18:32 | <annevk> | gsnedders, hmm, I hope mine is ok |
| 18:40 | <annevk> | painful how long all this takes |
| 18:52 | <jgraham> | annevk: The answer is "Yes I got it working" |
| 18:52 | <jgraham> | So it's not impossible :) |
| 18:53 | <annevk> | hah |
| 18:54 | <annevk> | did it take long? |
| 18:54 | <annevk> | seems to take ages here |
| 18:54 | <Hixie> | anything interesting going on in the space of web standards? |
| 18:54 | <jgraham> | I seems to have used libxml2-2.6.30 |
| 18:55 | <jgraham> | libxslt-1.1.22 |
| 18:56 | <annevk> | Hixie, apparently this one standard won't be finished until 2022, caused a lot commotion |
| 18:56 | <gsnedders> | Hixie: Umm, well, I haven't died yet. |
| 18:56 | jgraham | has found that installing stuff on webfactional was relatively easy compared to dreamhost |
| 18:57 | <gsnedders> | What needs optimizing: |
| 18:57 | <gsnedders> | sub.py:88(stringSubstitutions) |
| 18:57 | <gsnedders> | xref.py:65(addReferences) |
| 18:57 | <gsnedders> | outliner.py:56(build) |
| 18:57 | <gsnedders> | xref.py:101(getTerm) |
| 18:57 | <gsnedders> | Everything else within Anolis is quick enough |
| 18:58 | <jgraham> | gsnedders: Whilst that is indubitably a good thing I question whether it is strictly "in the space of web standards" |
| 18:58 | <gsnedders> | jgraham: Well, this is a space of web standards, and I'm here. |
| 19:00 | <annevk> | I guess I'm having more success with lxml now, as it's simply taking a long time to do something rather than reporting an error |
| 19:00 | <annevk> | though maybe it will fail in the end anyways... |
| 19:00 | jgraham | wishes he understood how latex worked so he didn't have to try random things untill one gave the right result |
| 19:00 | <jgraham> | s/latex/LaTeX/ |
| 19:01 | <annevk> | jgraham, do you remember lxml just running for a while without giving feedback? |
| 19:01 | <annevk> | (while installing) |
| 19:02 | <jgraham> | annevk: Don't remember. It might well take a non-trivial amount of time to compile though |
| 19:02 | <jgraham> | Have you tried seeting what's running using top or something |
| 19:03 | <annevk> | win win win import lxml works! |
| 19:03 | <annevk> | (just took a while) |
| 19:04 | <jgraham> | Now you just need to get mercurial running so that you can hg clone gsnedder's repo |
| 19:04 | <jgraham> | :) |
| 19:04 | <gsnedders> | :P |
| 19:04 | annevk | classifies that answer under "obvious fail" |
| 19:05 | <annevk> | s/answer/remark/ duh |
| 19:06 | <annevk> | jgraham, what is the best way to install html5lib ? |
| 19:06 | <gsnedders> | annevk: By doing a double backflip. |
| 19:07 | <annevk> | *sigh* |
| 19:08 | <annevk> | say I want to install it under opt/ as well |
| 19:09 | <jgraham> | annevk: svn co http://code.google.com/html5lib/whatver and then python setup.py install in the html5lib directory |
| 19:10 | <annevk> | jgraham, it will automatically move it around to the correct place? |
| 19:10 | annevk | never had an installed copy of html5lib |
| 19:11 | annevk | assumes it will be ok |
| 19:11 | <jgraham> | annevk: No. If you do python setup.py install --prefix=~ (might be --home) then it will put it somewhere in your home directory that is also on your sys.path. If you really want to install it somewhere different, look at the options under python setup.py --help |
| 19:11 | <jgraham> | as in it might be python setup.py --home=~ |
| 19:12 | <jgraham> | but prefix sounds more likely |
| 19:12 | <jgraham> | (it is likely to be put under ~/lib/python2.5/) |
| 19:15 | <annevk> | python runtests.py within anolis works |
| 19:15 | <annevk> | now I need some example Anolis code |
| 19:18 | annevk | finds out that part of Anolis has no documentation |
| 19:20 | <annevk> | gsnedders, in generator.py.process(), is **kwargs the commandline arguments you document elsewhere or something? |
| 19:23 | <met_> | figure microformat aka html5 figure http://microformats.org/discuss/mail/microformats-new/2008-September/001807.html |
| 19:25 | <annevk> | it would be better to just use HTML5... |
| 19:25 | <annevk> | but then browsers fail on <legend> :/ |
| 19:26 | <annevk> | gsnedders, if you can give me some sample code to work with that'd be great |
| 19:26 | annevk | off to buy some food |
| 19:30 | <met_> | annevk, for todays use it's ok, I wander why mf community doesn't prepare steps for future, like legend here, time instead of abbr there etc. |
| 19:31 | jgraham | still thinks that <legend> might be a deal breaker for <figure> |
| 19:32 | <hsivonen> | what's the problem that the figure microformat tries to solve? |
| 19:32 | <hsivonen> | that is, what's blocking prospective users from upgrading to HTML5? |
| 19:32 | <met_> | hsivonen, most probably the same problem as figure in HTML5 |
| 19:34 | <gsnedders> | annevk: See http://hg.gsnedders.com/anolis/file/cf4770338aa0/anolis and the dest arguments in the parser.add_option() calls |
| 19:35 | <met_> | hsivonen, users are ok, they can upgrade, but tools designed today shoud be ready for HTML5 already (tools can have long life), I wonder why microformats doesnt' say "this html5 syntax is 100% compatible with our syntax", so the tools can be ready for it |
| 19:35 | <Hixie> | jgraham: yeah, maybe. for <details>, too. i just really really don't want to have to introduce a 12th or 15th or whatever it is element for marking up headers/titles/captions/names |
| 19:36 | <jgraham> | Hixie: I know, but practicallity beats purity. There's no point in introducing the elements if no one can use them afterall. |
| 19:36 | <hsivonen> | met_: I also don't understand the motivation of asking for HTML4+ARIA validation |
| 19:37 | <hsivonen> | (a recent feature request for Validator.nu) |
| 19:37 | <Hixie> | jgraham: yeah. hopefully parsers can be fixed. |
| 19:42 | <jgraham> | Hixie: The problem with relying on fixing parsers is that the reptillian part of browser dev's brains which deals with their natural survival instincts tells them that changing parsers can break web compat. and so is bad ;) |
| 19:42 | <gsnedders> | hdh: http://hg.gsnedders.com/anolis/rev/bf926a2e75ab |
| 19:43 | <jgraham> | Hence parsers seem to have longer regeneration cycles than many aspects of browser code |
| 19:43 | <gsnedders> | hdh: (I think it was you who I said to bully me to do that if I forgot to, but you didn't so it took a while) |
| 19:43 | <Hixie> | jgraham: we're already relying on them changing the parser anyway |
| 19:43 | <Hixie> | jgraham: i'm happy to delay <figure> and <details> deployment until after the parser is deployed |
| 19:43 | <Hixie> | jgraham: they're not critical features |
| 19:46 | <jgraham> | Maybe not critical but <figure> is certianly something that various people have said to me "oh I've wanted that for a long time", so it is a bit annoying to have to go "oh well actually it won't work for the forseeable future because it's not backwards compatible" |
| 19:46 | <Hixie> | it's backwards compatible if people are willing to do tricks like have <span> elements around the place |
| 19:46 | <jgraham> | Especially if you've just been talking about how HTML 5 tries to preserve backwards compatibility |
| 19:47 | gsnedders | gives Hixie a hypocrite sticker |
| 19:47 | <Hixie> | backwards compatibility doesn't mean everything Just Works |
| 19:47 | <hdh> | gsnedders: it was me; thanks |
| 19:47 | <gsnedders> | then I run. |
| 19:47 | <Hixie> | i mean it's like saying the SQL database stuff "isn't backwards compatible" |
| 19:47 | <gsnedders> | It so is! |
| 19:48 | <jgraham> | Yeah, but stuff that is really new features is easier to explain than stuff that is just markup |
| 19:48 | <Hixie> | *shrug* |
| 19:49 | <jgraham> | Also, how can you sprinkle span to get FF to have the <legend> element in the DOM but no <fieldset>? |
| 19:49 | <Hixie> | <datalist> is no less of a new feature than <details> |
| 19:49 | <Hixie> | FF's <fieldset> thing is a bug and should be fixed regardless of any of this. |
| 19:50 | <gsnedders> | Oh, and can I bully anyone here into giving me comments on my personal statement at short notice? |
| 19:52 | <jgraham> | gsnedders: I can look this evening. By which I mean "in an hour or two" |
| 19:52 | <gsnedders> | jgraham: In an hour or two will do seeming I need to finish it first. |
| 19:52 | <gsnedders> | (Really, any time in the next 26 hours will do) |
| 19:54 | <jgraham> | gsnedders: OK, I can read it then |
| 19:54 | <Hixie> | ok. lunch time. then i'll find something else to delay working on <input>. |
| 19:54 | <annevk> | gsnedders, yeah, that sort of helps, but not really |
| 19:56 | <gsnedders> | annevk: What else do you want? |
| 19:57 | <annevk> | I would like a python API rather than a commandline API |
| 19:57 | <gsnedders> | It's basically the same |
| 19:58 | <annevk> | ok |
| 19:59 | <gsnedders> | annevk: See generator.fromFile and .fromToFile() |
| 19:59 | <gsnedders> | annevk: They take as keyword args those things |
| 20:00 | <annevk> | and how do I specify "those things" from Python? |
| 20:00 | <gsnedders> | generator.fromFile(StringIO(), lxml_html=True) |
| 20:01 | <annevk> | ok, so lxml_html=True, omit_optional_tags=False, ... |
| 20:01 | <annevk> | ? |
| 20:01 | <gsnedders> | yeah |
| 20:01 | <gsnedders> | w3c_compat_xref_a_placement=True |
| 20:01 | <gsnedders> | (HTML 5 needs that) |
| 20:04 | <annevk> | my generator file does not include fromFile btw |
| 20:05 | <gsnedders> | annevk: Get a later copy |
| 20:05 | <gsnedders> | :P |
| 20:05 | <annevk> | grmbl |
| 20:06 | <gsnedders> | http://hg.gsnedders.com/anolis/archive/tip.zip — or install mercurial and clone the repo |
| 20:07 | Philip` | installed Mercurial on Dreamhost, and it seemed to work alright |
| 20:07 | <Philip`> | (but it's easier to not do so) |
| 20:07 | jgraham | seems to have mercurial on DH also |
| 20:07 | gsnedders | has it on DH too |
| 20:07 | <jgraham> | I think I may have just tried my luck with easy_install mercurial |
| 20:11 | <annevk> | alright, that's slightly better |
| 20:12 | <annevk> | so I use fromFile and serialize the output as document which is then rendered in the browser |
| 20:12 | <annevk> | so to prevent xss it should be run on a separate domain |
| 20:12 | <gsnedders> | annevk: No, that doesn't help |
| 20:13 | <annevk> | what doesn't help? |
| 20:13 | <gsnedders> | annevk: They can put a URL in it that processes another doc in the spec gen |
| 20:13 | <gsnedders> | Actually, if you use POST, that's irrelevant |
| 20:13 | <gsnedders> | Hixie: But didn't you want GET? |
| 20:13 | <Philip`> | To prevent XSS, send it as text/plain with content-type-options:nosniff |
| 20:14 | <gsnedders> | How about application/octet+stream? |
| 20:14 | <annevk> | why not just use a separate domain? |
| 20:14 | <annevk> | i want to see the output the tool generates so i just need to hit refresh |
| 20:14 | <gsnedders> | annevk: Using POST or GET? |
| 20:14 | <annevk> | I don't see how that matters |
| 20:15 | <gsnedders> | annevk: If you use GET, I can use a URL that processes another document within it |
| 20:15 | <annevk> | that's not a XSS attack afaict |
| 20:16 | <annevk> | though it will probably use POST, as that makes the most sense for this |
| 20:17 | <annevk> | actually, ?url=http://www... would be a useful service too |
| 20:17 | <annevk> | guess both will be done |
| 20:18 | <Hixie> | gsnedders: i don't mind, i can do post from wget too iirc |
| 20:20 | <Hixie> | hsivonen: uri to "gez's table"? |
| 20:23 | <gsnedders> | Hixie: Was it you trying to run Anolis on Py2.3? |
| 20:23 | <Hixie> | yes |
| 20:23 | <gsnedders> | Hixie: Can you get http://hg.gsnedders.com/anolis/archive/tip.tar.bz2 and try again? |
| 20:24 | <Hixie> | i was just going to wait for anne or someone to set up a web service instead :-) |
| 20:24 | <gsnedders> | Hixie: I know, but I just want to see if it works on Py2.3 now :) |
| 20:25 | <Hixie> | give me a .sh file to run and i'll try it and give you the output |
| 20:25 | <Hixie> | include things like wget or whatever i need to fetch the file |
| 20:25 | <gsnedders> | Hixie: Just running python runtests.py |
| 20:25 | <gsnedders> | *run |
| 20:25 | <gsnedders> | :P |
| 20:26 | <Hixie> | wget http://hg.gsnedders.com/anolis/archive/tip.tar.bz2 |
| 20:26 | <Hixie> | tar jxvf tip.tar.bz2 |
| 20:26 | <gsnedders> | ah, you mean including getting it |
| 20:27 | <Hixie> | python anolis-eb6d50aa26f5/runtests.py |
| 20:27 | <Hixie> | Traceback (most recent call last): |
| 20:27 | <Hixie> | File "anolis-eb6d50aa26f5/runtests.py", line 27, in ? |
| 20:27 | <Hixie> | from anolislib import generator |
| 20:27 | <Hixie> | File "/home/.hardangervidda/ianh/temp/anolis/anolis-eb6d50aa26f5/anolislib/__init__.py", line 1, in ? |
| 20:27 | <Hixie> | from generator import * |
| 20:27 | <Hixie> | File "/home/.hardangervidda/ianh/temp/anolis/anolis-eb6d50aa26f5/anolislib/generator.py", line 22, in ? |
| 20:27 | <Hixie> | import html5lib |
| 20:27 | <Hixie> | ImportError: No module named html5lib |
| 20:27 | <gsnedders> | Hixie: Heh. At least I don't get a syntax error :) |
| 20:28 | <jgraham> | gsnedders: Trying to make it work on python 2.3 is a bad idea. Python 2.3 is missing some really nice things |
| 20:28 | <annevk> | either tonight or tomorrow i'll try to set it up |
| 20:28 | <gsnedders> | jgraham: reversed() was the only real thing missing |
| 20:28 | <annevk> | have to go now |
| 20:28 | <Hixie> | there's no rush |
| 20:28 | <Hixie> | i' deep in wf2 territory right now |
| 20:28 | gsnedders | is listening to Force Ten by Rush from Hold Your Fire |
| 20:28 | <gsnedders> | obviously there is. |
| 20:28 | <Hixie> | and don't want to do the switchover in the middle of the wf2 work |
| 20:28 | <gsnedders> | (bad joke, I know) |
| 20:29 | <jgraham> | gsnedders: You never use generators, or the codecs package or anything? |
| 20:29 | <gsnedders> | jgraham: Generators don't exist at all in 2.3? |
| 20:29 | <jgraham> | gsnedders: Not afaik |
| 20:30 | <gsnedders> | jgraham: They seem to, just not generator expressions. I removed the only one of those, as it made absolutely no difference to perf. |
| 20:30 | <jgraham> | Oh they were introduced in 2.3 |
| 20:31 | <jgraham> | Pre-2.3 only has a Python implementation of sets so html5lib may be significantly slower |
| 20:32 | <Hixie> | html5lib is too slow anyway :-) |
| 20:32 | jgraham | blames python |
| 20:33 | <jgraham> | Although clearly it is mainly my fault :) |
| 20:33 | <gsnedders> | jgraham: You chose the wrong language! |
| 20:33 | <Hixie> | pure python is just a really slow environment |
| 20:36 | <jgraham> | I wonder if mainstream python will get some of the dynamic speedup goodness that js is receiving |
| 20:38 | <hsivonen> | Hixie: http://juicystudio.com/wcag/tables/complexdatatable.html |
| 20:39 | <Hixie> | i already commented on that table and don't believe that nested headers would help it |
| 20:39 | <Hixie> | see the bugzilla bug |
| 20:40 | <Hixie> | http://www.w3.org/Bugs/Public/show_bug.cgi?id=5822 |
| 20:41 | <jgraham> | That table has been so overplayed. |
| 20:41 | <jgraham> | If there is one lesson we take from it, it should be taht requiring headers+id for such a simple layout reveals fundamental deisgn flaws |
| 20:42 | gsnedders | wonders if there was anything else that he'd been bullied into putting into Anolis 1.0 |
| 20:43 | <jgraham> | So, if we think it requires headers+id we would be better off fixing the fundamental problems rather than papering over the cracks |
| 20:44 | <Hixie> | gsnedders: cross-document cross-references? *puppy eyes* |
| 20:44 | <gsnedders> | Hixie: Even with puppy eyes, no. |
| 20:44 | <Hixie> | bummer |
| 20:44 | <Hixie> | see if i put it a <gsnedders> element now! :-P |
| 20:44 | <Hixie> | put in, even |
| 20:44 | <gsnedders> | :P |
| 20:45 | <gsnedders> | Hixie: What does that do? Make its content a hopeless romantic? |
| 20:45 | <Hixie> | it does nothing! |
| 20:45 | <Hixie> | :-P |
| 20:45 | <Hixie> | certainly doesn't add cross-references cross-document, anyway :-P |
| 20:45 | <gsnedders> | That's less than the sarcasm end tag does! |
| 20:45 | <gsnedders> | That at least makes you take a deep breath! |
| 20:46 | <Hixie> | hm on another note i guess i should get plane tickets to TPAC tonight |
| 20:47 | <cmlenz> | hey folks |
| 20:47 | <gsnedders> | I ought to decide what I'm doing for the weekend before |
| 20:48 | <gsnedders> | I need to decide whether to stay with my uncle for a few more days, or whether to come down to the south coast |
| 20:48 | <cmlenz> | just read http://blogs.msdn.com/ie/archive/2008/09/02/ie8-security-part-vi-beta-2-update.aspx and am wondering whether the "Restricting document.domain" example is what was intended by the spec |
| 20:48 | <cmlenz> | it's pretty weird that you can't reset document.domain to the original value |
| 20:49 | <cmlenz> | (note: I did not read through the comments, so I may have missed something :P ) |
| 20:50 | <Hixie> | http://www.whatwg.org/specs/web-apps/current-work/#domain |
| 20:50 | <hsivonen> | jgraham: I don't think it should require headers/id. but visually, it does have chained header cells |
| 20:51 | <Hixie> | visually that table is a mess |
| 20:51 | <cmlenz> | Hixie: read that, and in makes sense in general, but not in that specific example ;) |
| 20:51 | <cmlenz> | s/in/it |
| 20:51 | <Hixie> | cmlenz: what's the specific example that doesn't make sense? |
| 20:51 | <Hixie> | and what doesn't make sense about it? |
| 20:51 | <jgraham> | hsivonen: I agree |
| 20:52 | <jgraham> | I think chained headers should work |
| 20:52 | <cmlenz> | the one in the IE blog post |
| 20:52 | <cmlenz> | document.domain = "app1.example.com"; document.domain = "example.com"; document.domain = "app1.example.com"; |
| 20:52 | <cmlenz> | last one throws exception |
| 20:52 | <Hixie> | right |
| 20:52 | <Hixie> | is the spec unclear about this? |
| 20:52 | <jgraham> | At leat I think that untill someone shows me research that they are more confusing to end users than less information would be |
| 20:53 | <Hixie> | jgraham: that's backwards |
| 20:53 | <Hixie> | jgraham: i don't add features based on the absence of evidence that the features are harmful |
| 20:53 | <Hixie> | jgraham: i add them based on evidence that they are helpful |
| 20:53 | <cmlenz> | no, that follows the spec, I'm just saying it makes no sense that you can't set the property value back to the original, more restrictive, value |
| 20:53 | <Hixie> | cmlenz: aah |
| 20:53 | Hixie | points cmlenz to the /topic :-) |
| 20:54 | <cmlenz> | hmm :P |
| 20:54 | <Hixie> | that's just how most browsers behave |
| 20:54 | <Hixie> | and there are some half-sensible security reasons for it |
| 20:54 | <gsnedders> | (but only half) |
| 20:54 | <Hixie> | like, you wouldn't want to go user1.example.com -> example.com -> user2.example.com |
| 20:55 | <cmlenz> | sure, that rule is 100% needed |
| 20:55 | <cmlenz> | oh well, no big deal anyway, I guess |
| 20:56 | gsnedders | screams, "MY BROWSER! LET ME DECIDE HOW BIG I WANT THE WINDOW!" |
| 20:56 | <cmlenz> | just seems like something I'd be all WTF about when encountering it |
| 20:56 | <gsnedders> | cmlenz: Much of HTML is like that. |
| 20:56 | <cmlenz> | but then again, I probably never will |
| 20:57 | <hsivonen> | jgraham: what would you offer to authors who want to support legacy agents? |
| 20:57 | gsnedders | needs to get train tickets |
| 20:57 | <cmlenz> | maybe the spec could at least allow user agents to behave sensibly? :P |
| 20:58 | <jgraham> | hsivonen: For authors who want to support legacy agents, I would specify that headers pointing to td has to work, not be conforming |
| 20:59 | <jgraham> | Hixie: In this case you're arguing that providing only some of the information that a visual user has access to is more useful so I would say that you are making the more extraordinary claim |
| 21:00 | <gsnedders> | £21.50 to get from Lyon to Cannes. Not bad. |
| 21:00 | <jgraham> | (note I am more concerned with the normal case of a table marked up with <th> and <td> but no @headers than the accessibility-consultant-tweaked case where @headers is used) |
| 21:02 | <jgraham> | Hixie: I also think that if using only the first header in a chain is easier on the user then it is possible to implement that as a UA feature without breaking the spec, whereas I suspect at the moment all AT will just break the spec |
| 21:03 | <jgraham> | Since presenting less information in a new version is unlikely to go down well with the users |
| 21:03 | <gsnedders> | How long should I leave between my flight arriving at CDG and my train leaving? |
| 21:05 | jgraham | decides it is time to go home |
| 21:06 | <Hixie> | jgraham: the visual user has access to exactly the same information |
| 21:06 | <Hixie> | jgraham: if there are no headers and no scope attributes |
| 21:12 | <jgraham> | Hixie: Not really. If I have a table like <tr><th colspan="6">Sample numbers<tr><th>A<th>B[...]<tr><td>2.34<td>4.65 then the visual user can easilly see that the value 4.65 applies to "Sample number B", whereas the current headers scheme would iirc just show it applies to "B" |
| 21:12 | <jgraham> | Which is less information |
| 21:13 | <jgraham> | Now it may be that the UA wants to optimise away the display of that extra information |
| 21:14 | <jgraham> | By e.g. saying "Sample Number A 2.34, B 4.65" as the user moves across the table |
| 21:14 | <jgraham> | But it should still be associated by the algorithm so it is there if it is needed |
| 21:14 | <Hixie> | according to the spec as it stands today, unless i'm misreading it, 4.65 has two header cells associated with it |
| 21:15 | <Hixie> | Sample numbers and B |
| 21:15 | <jgraham> | Oh, wait a sec, the problem is that B doesn't have samle number as a header cell isn't it |
| 21:16 | <Hixie> | given what i've seen of users reading cells with screen readers, that doesn't seem like useful information |
| 21:16 | <jgraham> | OK, imagine a more complex example where a value could be from batch 1 or batch 2 |
| 21:16 | <Hixie> | i'd rather focus on real examples |
| 21:16 | <jgraham> | so the headers would be "sample", batch number sample number |
| 21:17 | <jgraham> | Hixie: So would I but making them up is easier :) |
| 21:17 | <Hixie> | it's easy to construct hypothetical tables that are nigh-on-impossible to understand even visually |
| 21:17 | <Hixie> | and that would have many nested levels of headers |
| 21:17 | <Hixie> | and god knows what else |
| 21:17 | <Hixie> | but real tables like that suck |
| 21:17 | <Hixie> | and should not be encouraged |
| 21:17 | <Hixie> | and people who want to make things accessible shouldn't just slap on some headers="" attributes to those tables |
| 21:18 | <Hixie> | because all they're doing is making it harder to show that their tables aren't accessible (even to visual users) |
| 21:18 | <jgraham> | I'm not suggesting that they just slap on @headers |
| 21:18 | <Hixie> | you're not, no |
| 21:19 | <jgraham> | I think it's weird if, in my example above there is no way to find out that "A" is a sample number rather than anything else |
| 21:19 | <hsivonen> | jgraham: why pointing to td? why not pointing to th? (re: legacy clients) |
| 21:19 | <jgraham> | Since that information is trivially avaliable visually |
| 21:19 | <Hixie> | jgraham: hit the "up" key |
| 21:20 | <jgraham> | hsivonen: Presumably legacy clients will continue to support legacy tables so de-facto @headers pointing to <th> will work anyway |
| 21:20 | <jgraham> | s/<th>/<td> |
| 21:20 | <jgraham> | We may as well spec reality |
| 21:21 | <Hixie> | reality is that nobody really uses headers="" and we could drop it and the only people who would notice are on public-html |
| 21:21 | <jgraham> | Hixie: Maybe. Doesn't work in some situations where there are data cells between the two headers; I'm sure Ben had a few examples like that |
| 21:22 | <jgraham> | (that was re "up key") |
| 21:22 | <Hixie> | i'm still confused as to the concept of headers with headers, personally |
| 21:22 | <jgraham> | Hixie: UAs will continue to support it whatever we way |
| 21:22 | <jgraham> | say |
| 21:24 | <jgraham> | Hixie: It just means that in cases where the table algorithm says C is a B is an A, it also says B is an A |
| 21:24 | jgraham | really is going home now |
| 21:26 | <Hixie> | would you ever focus a <label> and ask what <label>s it had? |
| 21:32 | <Philip`> | <label for=q>Secret question</label> <label id=q for=a><input><label> <label for=a>Secret answer</label> <input id=a> - you might want to go from the last input to its first label, and from that to the label's label, although that's probably a bit contrived :-p |
| 21:40 | <gsnedders> | Hmmm…¬ |
| 21:40 | <gsnedders> | s/¬// |
| 21:40 | <gsnedders> | I wonder if I could get anyone to pay me next summer to work on HTTP5 |
| 21:43 | <Hixie> | good luck |
| 21:43 | <Hixie> | though actually the market for paying people to work on standards is slightly better now than it was five years ago |
| 21:46 | <hsivonen> | gsnedders: try to disguise it as documentation for a SoC project |
| 21:48 | <gsnedders> | Ergh. I need 22p more to get free shipping on Amazon. Sux. |
| 21:48 | <Hixie> | how much is the shipping? |
| 21:49 | <Lachy> | I found out last night that I'm going to be speaking at PubCon in November this year. http://www.pubcon.com/bios/lachlan_hunt.htm |
| 21:50 | <Lachy> | I finally get a chance to go to the US :-) |
| 21:51 | <Hixie> | you'll get over your enthusiasm as soon as you arrive in the US airport :-) |
| 21:51 | <Hixie> | and have to go through hell :-) |
| 21:51 | <Hixie> | s/hell/airport security/ |
| 21:51 | <Hixie> | easy typo |
| 21:51 | <gsnedders> | What's the difference, again? |
| 21:52 | <Hixie> | hell doesn't lie about its intentions |
| 21:52 | <gsnedders> | Ah. |
| 21:52 | <hsivonen> | I think the enthusiasm stops before boarding the plane |
| 21:52 | <Hixie> | possible, depends where he's flying from |
| 21:53 | <hsivonen> | last year when going to TPAC, the first interrogation was in Amsterdam before boarding |
| 21:53 | <Lachy> | I will at least remain enthusiastic about temporarily leaving Norway |
| 21:53 | <Hixie> | amsterdam isn't the best place to leave from |
| 21:54 | <Hixie> | then again, if you're flying through amsterdam you'll be so distracted by the risk of losing your luggage in the changeover that you might not notice the interrogation |
| 21:54 | <gsnedders> | It's a better place to change than places like LHR and CDG |
| 21:54 | <Hixie> | oh well sure |
| 21:54 | <Hixie> | who would voluntarily fly through CDG |
| 21:54 | <Hixie> | jesus |
| 21:55 | <gsnedders> | I'm flying into CDG, but that's all right |
| 21:55 | <gsnedders> | Changing, however… |
| 21:55 | hsivonen | flies through CDG to TPAC |
| 21:55 | <Hixie> | the key to these interogations is to answer every question curtly and precisely without volunteering extra information and absolutely without showing any sign of irritation or annoyance |
| 21:55 | <gsnedders> | (I get straight off plane and on to train right next to the terminal to Lyon) |
| 21:55 | <Hixie> | and to do everything they tell you |
| 21:55 | <Lachy> | what's CDG? |
| 21:55 | <Hixie> | paris charles de gaule |
| 21:55 | <gsnedders> | Lachy: Aeroport de Charles de Gaulle |
| 21:55 | <gsnedders> | (près de Paris) |
| 21:56 | <Hixie> | a crazy retro-futuristic airport and, as far as i can tell, something that was originally only intended as a joke. |
| 21:57 | <hsivonen> | gsnedders: it seemed to me that going from CDG to Nice by train would have required swithing trains in paris with metro in between tha stations |
| 21:57 | <hsivonen> | but I may be wrong |
| 21:57 | <hsivonen> | the SNCF site suck |
| 21:57 | <hsivonen> | sites suck |
| 21:57 | <gsnedders> | hsivonen: No, you can get a train straight from CDG |
| 21:57 | <gsnedders> | If you have to change, it should be possible to do it somewhere sane like Lyon Part Dieu (where you'd just have to change platform) |
| 21:58 | <gsnedders> | Yeah, there are direct trains |
| 21:58 | <gsnedders> | Two per day |
| 21:58 | <gsnedders> | Certainly no need to go to Paris ever |
| 21:58 | <gsnedders> | (That's the one thing worse than changing planes at CDG) |
| 21:59 | <Hixie> | i was just thinking to myself that something was missing, and was trying to work out what it was. but i think i've worked it out. there hasn't been any crazy whining on public-html for days now. wtf. |
| 21:59 | <gsnedders> | Hixie: Want me to post something? |
| 21:59 | <Hixie> | no, no, i'm not complaining |
| 22:02 | <gsnedders> | hsivonen: In my experience flights within France are stupidly expensive |
| 22:02 | gsnedders | hopes the airline he's coming back on from NCE to EDI doesn't go bust |
| 22:02 | <gsnedders> | (Air France, which I'm flying out on, will just be bailed out by the government almost certainly, so there's no risk there) |
| 22:02 | <gsnedders> | (flying by Air France was 9p cheaper than EasyJet, oddly) |
| 22:02 | <gsnedders> | Now, I meant to do something useful this evening. I don't seem to have done so. |
| 22:06 | <hsivonen> | for the sake of the environment, I hope SNCF gets their Internet act together |
| 22:07 | <hsivonen> | (for some reason, db.de is the place that actually works for searching train schedules between France and Italy) |
| 22:08 | <gsnedders> | hsivonen: DB just generally has a good website that works, though |
| 22:08 | <gsnedders> | Well, not good in terms of design, but perfectly usable |
| 22:09 | <Hixie> | cff has a pretty good site too |
| 22:09 | <Hixie> | don't know if it goes into france |
| 22:10 | <gsnedders> | DB covers all of Europe, which is _really_ useful |
| 22:10 | <hsivonen> | too bad that you can't actually buy tickets there |
| 22:11 | <Hixie> | christ i can't wait to start using a spec gen that doesn't renumber all the IDs every time i add a new <dfn> |
| 22:11 | <gsnedders> | You can… within DE |
| 22:11 | <hsivonen> | but if you go to a vr.fi brick&mortar office, the person there uses the db.de extranet to book tickets e.g. between Hungary and Romania |
| 22:11 | <gsnedders> | Fun :\ |
| 22:13 | <Hixie> | <input> is coming along |
| 22:13 | <Hixie> | i wonder how long i can go without actually defining anything |
| 22:13 | <gsnedders> | Hixie: I'd guess at least the end of this month |
| 22:13 | <gsnedders> | Now, PRODUCTIVE TIME! |
| 22:15 | <gsnedders> | Hixie: Were you not born in Zürich? |
| 22:15 | <gsnedders> | Hixie: Which makes you calling it CFF interesting :) |
| 22:15 | <Hixie> | geneva |
| 22:15 | <gsnedders> | Ah, that makes more sense |
| 22:16 | Hixie | has a dozen cff trains and dozens and dozens of cff rolling stock |
| 22:16 | <Hixie> | swiss trains are so pretty |
| 22:16 | gsnedders | looks over his shoulder not remembering if he did that that one or not |
| 22:17 | <gsnedders> | No, I seem not to |
| 22:20 | <gsnedders> | I should upload my photos from Switzerland some time |
| 22:20 | gsnedders | notes he can't really label them because they're far too old |
| 22:20 | <gsnedders> | 2003, IIRC |
| 22:29 | <gsnedders> | I really suck at writing about why I want to do physics at uni |
| 22:32 | <gsnedders> | Meh. |
| 22:32 | <gsnedders> | I have around three times the amount about comp.sci. |
| 22:34 | gsnedders | hits the magic 4000 character limit |
| 22:42 | <annevk> | back, lets see if I can get some Python to run |
| 22:44 | <gsnedders> | annevk: You have 15 minutes to question me. |
| 22:44 | <gsnedders> | I will most likely ignore you anyway, as I need to get this done :) |
| 22:44 | <Hixie> | man, my abilites to procrastrinate know no bounds |
| 22:45 | <annevk> | oh, no questions, it probably won't work |
| 22:45 | <gsnedders> | Hixie: I've done nothing since I got home at 15:40 (it is now 22:48) |
| 22:45 | <annevk> | answers.com says that word is without the second r, Hixie |
| 22:45 | <Hixie> | wow, anne's pretty good at it too :-P |
| 22:45 | <Hixie> | (thanks) |
| 22:45 | <annevk> | :p |
| 22:46 | <Hixie> | gsnedders: the thing is i've done lots |
| 22:46 | <gsnedders> | Hixie: I've been trying to write personal statement for UCAS since then, with the exception of 15 mins for supper |
| 22:46 | <Hixie> | gsnedders: just not what i set out to do! |
| 22:46 | <gsnedders> | Oh, I've done plenty |
| 22:46 | <Hixie> | just look at the huge table in the <input> section |
| 22:46 | <gsnedders> | I've been looking up stuff for October, and photographic stuff |
| 22:46 | <Hixie> | yet still nothing is defined |
| 22:46 | <gsnedders> | Oh, and I've been keeping up with most email as it comes in |
| 22:46 | <gsnedders> | And been active in IRC the whole time |
| 22:46 | <Hixie> | sadly there has been so little e-mail |
| 22:46 | <gsnedders> | Not good when UCAS form needs to be done soon. |
| 22:47 | <annevk> | ah, finally, Hixie replaced abuse of <em> with <i> |
| 22:47 | annevk | reads diffs first |
| 22:47 | <Hixie> | annevk: yet more example of my procrastination! |
| 22:48 | gsnedders | reads diff too |
| 22:48 | <gsnedders> | MUST WORK. |
| 22:48 | gsnedders | tweets about our ability to procrastinate |
| 22:49 | <Hixie> | i'd volunteer to ban you for 20 minutes at a time to help you work |
| 22:49 | <Hixie> | but we don't have ops :-) |
| 22:49 | <gsnedders> | hah. |
| 22:49 | <gsnedders> | Hixie: I'd just go to other channels. |
| 22:49 | <Hixie> | d'oh |
| 22:49 | <gsnedders> | http://twitter.com/gsnedders/statuses/923871883 |
| 22:49 | <annevk> | file "A label and a button" could also be a "A text field and a button" ... |
| 22:50 | <Hixie> | annevk: i'd rather not encourage people to give a text field, for security reasons |
| 22:51 | gsnedders | passes Hixie a text field |
| 22:52 | Hixie | gets an image of a crop field growing text |
| 22:54 | <gsnedders> | Hixie: Could I make you useful and get you to read over my personal statemnet? |
| 22:54 | <gsnedders> | *statement |
| 22:54 | <Hixie> | you could try |
| 22:54 | <gsnedders> | :) |
| 22:54 | <Hixie> | i make no promises |
| 22:54 | <gsnedders> | You sensible person. |
| 22:54 | <gsnedders> | I'm meant to be doing far too much because of it. |
| 22:55 | <gsnedders> | http://www.heas.com.au/publications/images/idealperformancestate.gif — I'm too far right on that. |
| 22:57 | <gsnedders> | Now, to write a conclusion. |
| 23:03 | <Hixie> | oooooh |
| 23:03 | <Hixie> | we're two weeks away from HTML5 being in CR according to the W3C timetable |
| 23:03 | <annevk> | </sarcasm> |
| 23:06 | <jgraham> | gsnedders: I'm around again fwiw |
| 23:06 | <gsnedders> | jgraham: email coming soon |
| 23:07 | <gsnedders> | (like, I'm writing it now) |
| 23:07 | <gsnedders> | (the email, not the personal statement) |
| 23:16 | <gsnedders> | sent |
| 23:17 | <annevk> | sigh |
| 23:17 | <annevk> | from anolislib import generator gives a 500 |
| 23:18 | <jgraham> | annevk: What's in the log? |
| 23:23 | <Hixie> | what's the point of a personal statement again? :-) |
| 23:23 | <jgraham> | gsnedders: I'll send more detailed comments by email but my first impressions are a) you need to put computer science second and physics first b) you mention things that are bad too often |
| 23:24 | <jgraham> | (e.g. talking about the simplepie codebase being messy) |
| 23:24 | <Hixie> | my first thought was "these paragraphs are too long" but i'm not sure how to evaluate it beyond that given my lack of recall of what a personal statement is |
| 23:24 | <annevk> | oops, the first line didn't point to my custom version of python |
| 23:24 | <jgraham> | Hixie: The point is is to seel yourself to the university as someone who is interested in the subject and so likely to turn up to lectures |
| 23:25 | <jgraham> | s/seel/sell |
| 23:25 | <Hixie> | ah |
| 23:25 | <Hixie> | "I love studying" seems like a theme to focus on then |
| 23:25 | <annevk> | the error log does get an entry for every refresh though, "UserWarning: This version of libxml2 has a known XPath bug. Use it at your own risk." |
| 23:25 | <annevk> | sigh |
| 23:26 | <jgraham> | (I guess there is more to it than that but they want to make sure that they don't take people who chose the course at random) |
| 23:26 | annevk | will ignore useless warnings for now |
| 23:27 | <Hixie> | gsnedders: send me a copy again when it's fixed jgraham's pretty overarching feedback and has smaller paragraphs :-) |
| 23:43 | <annevk> | and I was wondering why file upload didn't work, as it happens specifying enctype=multipart/form-data is actually important |
| 23:44 | <annevk> | would be nice if that defaulted to the appropriate type depending on whether or not <input type=file> is present but I guess it's too late now |
| 23:57 | <dglazkov> | Hixie: remember this cool doogie-bob concept of a JS "promise" that you once described here? |
| 23:57 | <dglazkov> | I want to remember how it worked |
| 23:58 | <dglazkov> | this had to do with async calls |
