| 00:01 | <Philip`> | Is html5.org down? |
| 00:09 | <hasather> | Philip`: this might be it: http://www.dreamhoststatus.com/2007/12/15/annoying-network-issues/ |
| 00:09 | <jgraham_> | Philip`: It seems to be down for me as weel |
| 00:09 | <jgraham_> | s/weel/well/ |
| 00:14 | <Philip`> | "An OrientationInterpolator interpolates between two orientations by computing the shortest path on the unit sphere between the two orientations. The interpolation is linear in arc length along this path." - hmm, it'd be much more obvious if they simply said to use quaternion slerp, since I wouldn't have to work out whether that's what they actually mean |
| 00:14 | <jacobolus> | Hixie: sorry to keep bugging you. but any thoughts about that draft? |
| 00:15 | <jacobolus> | (http://pastie.textmate.org/private/xsvmjyopvftavgbcbjpexa) |
| 00:39 | <Philip`> | Hooray, a shiny monkey head on a declaratively animated rotating cube |
| 06:38 | <Hixie> | http://groups.google.com/group/show-in-a-box/browse_thread/thread/2e53cb4c3df04578/b196aaf9676688fe?hl=en&q=html5#b196aaf9676688fe needs some subscribe-and-comment love |
| 06:38 | <Hixie> | if anyone's up for it |
| 06:40 | <Hixie> | and if anyone wants to explain to this guy that DRM really isn't part of the problem, be my guest. He event cites me but apparently believes my summary isn't an accurate representation of the situation. http://www.sourfizz.com/2007/12/15/html5-working-group-rejects-open-media-formats/ |
| 07:41 | <doublec> | I've subscribed to the google groups one and will reply |
| 08:22 | <hsivonen> | whoa. MathML 3.0 |
| 08:23 | <hsivonen> | aaargh. they still use entities an *redefine* a bunch of them |
| 08:25 | <hsivonen> | hmm. the MathML 3.0 draft uses some mighty weird bolding in code. I wonder if the weirdness comes from TeX4ht or similar |
| 08:33 | <Hixie> | doublec: cool, thanks |
| 10:37 | <Philip`> | Is someone able to delete wiki pages, like http://wiki.whatwg.org/wiki/Help:Contents and http://wiki.whatwg.org/wiki/Brooklyn_indie_market ? |
| 10:39 | <anne-mac> | in theory |
| 10:41 | Philip` | assumes it's limited to sysops |
| 10:42 | <Philip`> | Aha, thanks |
| 10:59 | <anne-mac> | you should ask lachy or so to get you upgraded to sysops |
| 12:15 | <Philip`> | index.php?pid=2&&opcje=a:1:{i:0;s:6:"wyszuk";} |
| 12:16 | <Philip`> | What kind of syntax is that? It looks somewhat reminiscent of JSON, but it isn't |
| 12:59 | <inimino> | Philip`: that's from PHP |
| 12:59 | <tndH> | yeah, it's output from serialize() |
| 12:59 | <inimino> | right |
| 13:01 | <Philip`> | Ah, thanks |
| 13:56 | <anne-mac> | we need to update http://www.whatwg.org/specs/ ... |
| 13:58 | <gsnedders> | what a bizarre bug in Opera. |
| 13:59 | gsnedders | looks at XMLHttpRequest |
| 13:59 | gsnedders | notices other stuff |
| 14:00 | <gsnedders> | anne-mac: AFAIK nothing actually checks getreponseheader() matches field-name |
| 14:00 | <Dashiva> | gsnedders: Bizarre bugs are the best kind |
| 14:00 | <Philip`> | Does anybody use XPointer? |
| 14:00 | <Dashiva> | Like when setting a className on a table cell causes it to quadruple in width, even though the class doesn't exist |
| 14:01 | <hdh> | me, with this fx extension http://www.codedread.com/fxpointer/ |
| 14:02 | <gsnedders> | Dashiva: ah. the bug I found is considerably less bizarre. |
| 14:02 | <gsnedders> | Dashiva: XMLHttpRequest.getResponseHeader("connection") always returns "" in Opera |
| 14:03 | <Dashiva> | There was also that one where innerHTML of responseXML had no closing tags |
| 14:04 | <Dashiva> | hm |
| 14:04 | <Philip`> | hdh: Ah, that looks potentially useful |
| 14:05 | Philip` | is planning to use XPointer for an extension of X3D so you can include bits of uncooperative external X3D files |
| 14:05 | <Dashiva> | gsnedders: Must be some special condition you've found, it works in the general case. Got a testcase? |
| 14:05 | <Philip`> | (since the normal X3D inclusion mechanism seems to require cooperation and explicit exports, which is a pain) |
| 14:06 | <gsnedders> | Dashiva: <http://hg.gsnedders.com/cgi-bin/hgwebdir.cgi/http-parsing/file/29a9ec814634/tests/response/>; — I need to make an actual test-case though |
| 14:08 | <hdh> | http://blog.codedread.com/archives/2007/10/30/fxpointer-link-exactly/ has a link to an userjs version for opera |
| 14:10 | <Philip`> | Hmm, looks like that only supports the element(confusing and fragile child node numbering) system |
| 14:10 | <Philip`> | whereas I want XPath because that's more fun |
| 14:11 | <Philip`> | (and XPath is supported by all the browsers I care about) |
| 14:11 | <gsnedders> | Dashiva: it may be only when the only header is a Connection: Close |
| 14:16 | <hdh> | the userjs demo has an xpath, #xmlns(e=http://www.w3.org/1999/xhtml)xpointer(//e:li[4]) |
| 14:18 | <Philip`> | hdh: Oh, oops, I was looking at the wrong .js file on that page |
| 14:20 | Philip` | will try copying some of that code |
| 14:28 | <gsnedders> | http://http-parsing.gsnedders.com/#date-formats makes sense now I see http://support.microsoft.com/kb/234067 |
| 14:45 | <gsnedders> | hmmm. |
| 14:51 | <gsnedders> | even odder things happen when you have the developer console open in Opera |
| 14:52 | <gsnedders> | and if there are other headers too |
| 14:59 | <Philip`> | Odd things happening in a web browser? That's unfortunately unsurprising :-( |
| 15:09 | gsnedders | has reported the bug to Opera Software, but has been unable to write a minimalistic test case, so is just linking to the revision of the test suite |
| 15:11 | gsnedders | writes in an email "specs be damned" |
| 15:14 | Philip` | wonders if that email is about braces not being allowed in URIs |
| 15:14 | <gsnedders> | yeah |
| 15:16 | Philip` | wonders if he should look for braces in other URI attributes |
| 15:17 | Philip` | wonders if anyone has a list of what attributes take URIs |
| 15:17 | <gsnedders> | another way to make sure a spec never gets implemented by UAs: require HTML is to be parsed by SGML parsers. |
| 15:17 | <gsnedders> | Philip`: I tend to just search through the HTML 4.01 DTD for %uri |
| 15:17 | <gsnedders> | (which misses out non-conforming tags/attributes, but hey) |
| 15:20 | <Philip`> | http://www.cl.cam.ac.uk/~pjt47/misc/uris-with-braces.txt |
| 15:21 | <Philip`> | Oops |
| 15:21 | Philip` | fixes minor bug |
| 15:23 | <Philip`> | "If RFC 3986 defined what to do with non-conformant URIs" ... and if browsers implemented those requirements perfectly |
| 15:26 | <gsnedders> | :P |
| 15:27 | <Dashiva> | I don't think it would work even then |
| 15:28 | <Dashiva> | The issue is on the markup level, so either the UAs let it work (current situation), or they refuse to accept those URLs (not going to happen) |
| 15:30 | <gsnedders> | Dashiva: well, it depends what the defined handling was anyway |
| 15:31 | <hsivonen> | putting GUIDs in URIs strikes me as awful URI design |
| 15:32 | <Philip`> | I think clean URI design comes quite low on the prioritised list of problems that many authors have |
| 15:33 | <gsnedders> | "No, they rely on user agents not checking for syntactically correct URIs. They aren't using URIs." — reply to me |
| 15:35 | <Philip`> | Are they using syntactically incorrect URIs, or are such things defined to not exist? |
| 15:35 | <gsnedders> | I guess they don't exist. |
| 15:35 | <gsnedders> | (technically, if you take RFC 3986 literally, a string is either a conforming URI or it isn't — nothing else) |
| 15:35 | <gsnedders> | *it isn't a URI whatsoever |
| 15:36 | <Philip`> | It does sound sensible for other specs (like HTML) to define how non-conforming URIs are handled, since different cases will prefer it to be handled in different ways |
| 15:36 | <gsnedders> | WebKit for one handles it all in one place, regardless of where it has come from |
| 15:37 | <gsnedders> | Under the same argument HTML error handling should be defined by other specs (like HTTP) |
| 15:37 | <Philip`> | e.g. the URI RFC can't say that clicking on an HTML link with an invalid URI should log an error message and perform no action, because it's not its job to define anything about HTML |
| 15:38 | <Dashiva> | It seems like two different perspectives. One says "If it doesn't follow the rules, it's not an URI, who cares". The other says "This was used as an URI, but doesn't follow the rules, what do we do?" |
| 15:38 | <gsnedders> | Philip`: It can say to strip leading/trailing whitespace, and escape any other invalid characters as pct-encoded |
| 15:38 | <gsnedders> | Philip`: which is what is used everywhere, AFAIK |
| 15:39 | <Philip`> | "everywhere" does not include HttpClient, it seems |
| 15:39 | <Philip`> | Invalid uri 'http://www.springer.com/east/home?SGWID=5-102-70-1033334-detailsPage=journal|description|description';: Invalid query |
| 15:39 | <Philip`> | (is what it complains about) |
| 15:40 | <gsnedders> | s/everywhere/WebKit\/Gecko\/Trident\/Presto/ then |
| 15:40 | <Philip`> | Elsewhere: "GET failed: HTTP/1.1 508 unused" - huh? |
| 15:41 | gsnedders | headdesks |
| 15:43 | <Philip`> | java.lang.IllegalArgumentException: Invalid uri 'http://www.metmuseum.org/special/se_event.asp?OccurrenceId={A238C1BA-B848-11D3-936D-00902786BF44}';: Invalid query |
| 15:43 | <Philip`> | Ah, HttpClient does complain about braces too |
| 15:45 | <Philip`> | Quite a few invalid redirects too |
| 15:45 | <Philip`> | Error on http://gutenberg.spiegel.de/autoren/dauthend.htm : Invalid redirect location: http://gutenberg.spiegel.de/?id=19&autor=Dauthendey, Max&a |
| 15:45 | <Philip`> | utor_vorname= Max&autor_nachname=Dauthendey |
| 15:45 | <Philip`> | (...minus the linebreak) |
| 15:51 | <gsnedders> | more silly arguments on digg. |
| 15:51 | <gsnedders> | equal crime rates in a place where guns are illegal and one where they aren't, therefore you're safer if you have a gun. |
| 15:52 | <hdh> | just ban bullets and let people use their metal butts |
| 15:53 | <gsnedders> | That's the Swiss solution, IIRC |
| 15:53 | <Philip`> | If guns are banned, more bank robbers will have to use bananas instead, which are a good source of potassium, so everyone wins |
| 15:53 | <gsnedders> | (everybody who has done national service there has a gun, but the ammunition is sealed — the prison sentences for breaking that seal are c. 20 years IIRC) |
| 16:26 | <hdh> | POST is non-idempotent, so why don't input[type=button][method=POST] have a pushed state like a:visited? |
| 16:29 | <Philip`> | hdh: Shouldn't it be the other way around? Since POST is non-idempotent, it's different every time you click the button, whereas :visited is meant for things that are the same as what you've seen before |
| 16:32 | <hdh> | oh yes |
| 17:27 | <hdh> | what's the progress' "denominator punctuation character"? |
| 17:28 | <gsnedders> | hdh: follow the reference to "steps for finding one or two numbers of a ratio in a string" |
| 17:28 | <G0k> | hey all |
| 17:28 | <G0k> | args |
| 17:28 | <hdh> | gsnedders: thanks |
| 17:28 | <gsnedders> | G0k: see you like to stay around |
| 17:29 | <G0k> | fucking colloquy |
| 17:29 | <G0k> | it lets me join a channel |
| 17:29 | <G0k> | then forgets to echo anything in it |
| 17:29 | <G0k> | so i have to rejoin |
| 17:29 | <gsnedders> | no, it does output stuff into it |
| 17:29 | <gsnedders> | the bug is in WebKit :P |
| 17:29 | <G0k> | how does it work sometimes but not other times? |
| 17:29 | <G0k> | and is there some other workaround? |
| 17:29 | <gsnedders> | G0k: Clear the window |
| 17:30 | <gsnedders> | G0k: shortcut is cmd + k, no idea what menu |
| 17:30 | <G0k> | cmd+k? |
| 17:30 | <G0k> | ah |
| 17:30 | <G0k> | yeah |
| 17:30 | <G0k> | so i have an idea wrt server sent events |
| 17:31 | <gsnedders> | http://bugs.webkit.org/show_bug.cgi?id=14171 |
| 17:31 | <G0k> | ah |
| 17:31 | <G0k> | well. poop. |
| 17:31 | <G0k> | anyway |
| 17:32 | <G0k> | so 90% of the proposed uses of server sent events has been for chat-style stuff |
| 17:32 | <G0k> | so rather than invent that new dom event stream protocol |
| 17:32 | <G0k> | how about we just use XMPP? |
| 17:33 | <gsnedders> | G0k: and add DOM hooks to it? |
| 17:34 | <G0k> | well i mean you could send dom events as the payload |
| 17:34 | <G0k> | or just use (for example) the messaging and presence functionality already in it |
| 17:38 | <Philip`> | Hmm, using X3D inline in XHTML doesn't actually seem like a stupid idea any more, when I can import stuff like meshes from external files |
| 17:42 | <hdh> | and some nice CPU; I can't try x3d out, nobody builds for 64bit gnu/linux |
| 17:43 | <Philip`> | hdh: Do you mean normal X3D plugins, or Mozilla's canvas-3d extension? |
| 17:44 | <hdh> | I mean the two builds on http://philip.html5.org/demos/canvas/3d/x3d/test.html |
| 17:45 | <Philip`> | Ah, okay |
| 17:45 | <Philip`> | (I've just been running on Windows, since I don't have useful graphics hardware on my Linux computer) |
| 17:46 | <G0k> | we can combine 3D canvases with server sent evetns |
| 17:46 | <hdh> | quake clone? |
| 17:46 | <Philip`> | I would assume the Mozilla extension could be built, but Mozilla is not the quickest thing to compile |
| 17:47 | <Philip`> | I've already done an online multiplayer FPS using Canvex :-) |
| 17:47 | <zcorpan_lap> | Philip`: pointer? |
| 17:47 | <hdh> | I got the video-enabled fx3 here; I'll see about the plugin |
| 17:47 | <G0k> | yeah 3d games ftw |
| 17:48 | <Philip`> | (using pipelined XHR (in Opera) and dynamically-inserted <script>s (in Firefox) to get improved latency) |
| 17:48 | <Philip`> | zcorpan_lap: It's at file:/// somewhere, unfortunately |
| 17:56 | <zcorpan_lap> | Philip`: ok |
| 18:02 | Philip` | updates his code |
| 18:03 | <Philip`> | Now you can export some static meshes from e.g. Blender, and then write something like http://philip.html5.org/demos/canvas/3d/x3d/scene.x3d to combine them and animate them, which seems useful |
| 18:05 | <Philip`> | I think that mostly addresses my concerns over how anyone could make complex scenes, which have too much mesh data to be editable in a text editor and too much complex XML linking to edit in a 3D editor |
| 18:06 | <Philip`> | though now I need to find some way to import an external mesh and then inject some extra nodes into it, like replacing its <Material> |
| 18:07 | Philip` | guesses XSLT might be too complex |
| 18:51 | <Philip`> | Oh, maybe XSLT isn't that bad after all |
| 19:57 | <Dashiva> | So the spec says that whitespace "after body" should be treated as "in body", and that whitespace in "trailing end phase" should be treated as in "main phase", which becomes "in body". |
| 19:57 | <Dashiva> | "in body" says "Append the character to the current node" |
| 19:58 | <Dashiva> | So if you have whitespace between </body> and </html>, and after </html>, is that one or two whitespace text nodes at the end of body? |
| 20:28 | <webben> | Is there no way to hide a machine-friendly version of a human-friendly duration in HTML5 as currently drafted? |
| 20:29 | <webben> | e.g. in XHTML 2 as currently drafted one could do: <span content="PT2M43S">almost three minutes</span> |
| 20:30 | <Dashiva> | Isn't that what <time> does? |
| 20:30 | <Philip`> | <time> seems to only do absolute times |
| 20:30 | <webben> | A duration is not a datetime. |
| 20:31 | <Philip`> | Is http://www.whatwg.org/specs/web-apps/current-work/multipage/section-common1.html#time-offsets relevant? |
| 20:32 | <webben> | Philip`: If there were a DURATION with a VALUE attribute, it might be. Otherwise, it's just a way of parsing such values. |
| 20:35 | <webben> | The context for this is I'm wondering how hAudio durations could be correctly represented in HTML5. This problem of hiding machine-friendly versions of human-friendly data is just going to recur again and again. |
| 20:36 | <Philip`> | Does hAudio need machine-friendly vs human-friendly versions? I would assume something like "5:02" or "1:58:21" would be easy for both machines and humans |
| 20:38 | Philip` | doesn't know if there are i18n concerns with writing times in that format |
| 20:38 | <webben> | Philip`: Funnily enough, that's the argument I've just put to the list (so I'll wait and see counterexamples or complaints from parser implementers). There is certainly one case where authors might want to not use a format like that: when you're describing a piece of audio in a paragraph of extended descriptions. |
| 20:38 | <webben> | *description. |
| 20:39 | <webben> | e.g. "and the tedious concerto lasts <duration value="1:04:24">a whole hour</duration>! |
| 20:40 | <webben> | My argument is that the i18n problems in this particular case probably don't justify hacking around in HTML 4.01, but it would be nice to have a clean way to represent such things in HTML5. |
| 20:41 | <webben> | e.g. if you look at http://www.top100.cn/Product/Product.aspx?productid=S0093820000 Chinese users are apparently happy with arabic numerals in minutes and seconds for durations. |
| 20:44 | <Philip`> | Looks like there's not as much variation as with dates/times, which is nice |
| 20:44 | Philip` | sees that page has "2007-3-5" |
| 23:10 | <Lachy> | is anyone else having trouble connecting to irc.w3.org? |
| 23:10 | <gsnedders> | I'm on it fine |
| 23:10 | <Lachy> | weird. I'm getting unknown host errors |
| 23:11 | <gsnedders> | just disconnected and reconnected without issue |
| 23:13 | <Lachy> | hmm. now it's working |
| 23:14 | <csarven> | 6665 is okay for me |