| 07:18 | <maikmerten> | uh, on media elements... when is MEDIA_ERR_DECODE supposed to be set? As soon as the client realizes that he can't playback the media (e.g. no source with a fitting media codec)? This would could be during page load. Or is the error supposed to be set when actually attempting playback? |
| 07:21 | <maikmerten> | This is somewhat important so scripts can replace non-fitting (for the content) media elements with e.g. a video plugin fallback |
| 08:27 | <hsivonen> | Lachy: I think NonCommercial is more devious than NoDerivatives |
| 08:29 | <hsivonen> | http://twitter.com/karlw3c/statuses/906253637 |
| 08:34 | <hsivonen> | zcorpan: fixed the popup to be consistent with the field |
| 08:43 | <hsivonen> | zcorpan: I added aria-labelledby (but am not competent to test the result) thanks |
| 08:44 | <Hixie> | what should i call the insertion mode to replace the generic cdata/rcdata algorithm? |
| 08:44 | <Hixie> | "in text block" is all i have come up with so far |
| 08:44 | <hsivonen> | 'in [R]CDATA' |
| 08:45 | <Hixie> | is "in CDATA/RCDATA" ok? |
| 08:45 | <hsivonen> | yes |
| 08:45 | <Hixie> | ok |
| 08:45 | <Hixie> | thanks |
| 08:45 | <annevk> | that or (R)CDATA |
| 08:45 | <Hixie> | yeah i wanted to avoid parentheses everywhere |
| 08:45 | <Hixie> | it would look weird i think |
| 08:45 | <annevk> | fair enough |
| 08:47 | <Hixie> | crap i already have something called the "secondary insertion mode" |
| 08:47 | <Hixie> | state machines really don't handle push/pop "subroutine" semantics well |
| 08:48 | <annevk> | I guess http://twitter.com/karlw3c/statuses/906253637 was inevitable |
| 08:49 | <annevk> | though the reasoning is somewhat flawed: http://twitter.com/karlw3c/statuses/906259528 |
| 08:50 | <hsivonen> | Hixie: can we please have a flag instead of secondary modes |
| 08:51 | <Hixie> | you want to check a flag with every single token instead of jsut dispatching on a mode? |
| 08:51 | <Hixie> | that seems... bad |
| 08:53 | <hsivonen> | Hixie: it's only needed once per character *run* and once per end tag |
| 08:55 | <roc> | that's not flawed, that's just delusional |
| 08:56 | <roc> | At a wild guess, seamless sandboxed iframes would be quite hard to implement in Chrome |
| 09:01 | <hsivonen> | Hixie: are you expecting the perf characteristics of a mode to be better than a flag because your runtime environment makes function pointer-based calls faster than conditional branch on flag? |
| 09:06 | <annevk> | Hixie, when will annotate IDL fragments with Null=Empty and all? |
| 09:11 | <Hixie> | hsivonen: that seems like more than necessary. it also wouldn't really fit into the way the spec is written. |
| 09:12 | <Hixie> | hsivonen: i'm expecting C++ to make an extra case in a switch statement to be cheaper than an additional if when the switch still has to be there |
| 09:12 | <hsivonen> | Hixie: yes |
| 09:13 | <hsivonen> | Hixie: but if you have a secondary mode, switching modes in mid-switch sucks |
| 09:13 | <Hixie> | annevk: i'm waiting for heycam to update the spec before looking at idl |
| 09:13 | <hsivonen> | Hixie: which is why I have a separate outer switch for foreign content |
| 09:13 | <Hixie> | annevk: then we're going to need a lot of research |
| 09:13 | <Hixie> | hsivonen: ? |
| 09:13 | <hsivonen> | it's a switch instead of an 'if' to make fallthrough work |
| 09:13 | <annevk> | Hixie, lots of trivial research :) |
| 09:14 | <hsivonen> | Hixie: so if a switch on mode |
| 09:14 | <Hixie> | hsivonen: i think you're making assumptions about your implementation that the spec doesn't |
| 09:14 | <hsivonen> | Hixie: and you tell me to do something according to the secondary mode |
| 09:14 | <hsivonen> | Hixie: I'd need to re-enter the switch with a new condition |
| 09:15 | <hsivonen> | Hixie: so having secondary modes is not good |
| 09:15 | <hsivonen> | unless they are a latent stack |
| 09:15 | <hsivonen> | and you never dispatch on the secondary mode |
| 09:15 | <Hixie> | hsivonen: for foreign stuff maybe, but for this rcdata/cdata thing there's no fallback |
| 09:16 | <hsivonen> | Hixie: ok. well that's cool then |
| 09:16 | <Hixie> | it's just two options: text=>append, end tag=> do whatever is appropriate for that end tag, and switch back to the old mode |
| 09:18 | <Lachy> | hsivonen, IMHO, the worse CC licences are, in order, ND, NC, and SA. The reason is that ND prevents anyone from actually doing anything, except republishing unmodified copies which isn't very creative, and SA is bad because of the viral effect of copyleft licences |
| 09:18 | <Hixie> | SA is good because of the viral effect of copyleft licenses :-) |
| 09:18 | <Hixie> | still, having yet another thing called "secondary mode" isn't an option |
| 09:19 | <Hixie> | and having one thing called secondary and another called alternative or something would be weird |
| 09:21 | <Lachy> | SA is good because it prevents proprietary use, but it's also bad because it affects the ability to use it in more permissive, non-copyleft licences without making the whole thing copyleft. |
| 09:21 | <hsivonen> | Lachy: at least ND is upfront about not letting you do stuff. NC is a trap. First you do something and then find that you are barred from doing stuff that touches money. |
| 09:21 | <annevk> | just call it RCDATA/CDATA mode |
| 09:21 | <hsivonen> | copyleft doesn't forbid commercial use |
| 09:21 | <annevk> | oh, nm |
| 09:22 | <Lachy> | ok, fine. I'll say they're equally bad |
| 09:22 | <hsivonen> | Lachy: even RMS thinks NDness is OK for opinion writings |
| 09:23 | <Hixie> | Lachy: SA->SA->SA->... is better than CC->CC->proprietary |
| 09:23 | <hsivonen> | in fact, RMS' essays are under a one-sentence formulation of CC-by-nd |
| 09:23 | <Lachy> | so? Is that an appeal to authority? |
| 09:24 | <Hixie> | if one finds oneself having a position even more extreme than RMS, one is usually wrong :-) |
| 09:24 | <Hixie> | RMS is the definition of the edge of the spectrum here :-P |
| 09:25 | <hsivonen> | Lachy: what Hixie said |
| 09:25 | <Lachy> | the spectrum isn't linear |
| 09:26 | <Hixie> | an "edge" exists around any finite phase space |
| 09:26 | <Lachy> | I don't have a big problem with BSD type licences because they allow non-copyleft proprietary uses, whereas RMS does |
| 09:27 | <hsivonen> | Lachy: NC effectively puts you into this situation: http://overlawyered.com/2006/10/why-there-arent-dvds-of-some-of-your-favorite-old-tv-series/ |
| 09:27 | <Lachy> | but I have a problem with my own stuff being forced into using a copyleft licence if I want to make use of copyleft licenced stuff |
| 09:27 | <Hixie> | i guess i'm going to have an "original insertion mode" for rcdata/cdata and a "secondary insertion mode" for foreign |
| 09:27 | <Dashiva> | Maybe call it 'foreign insertion mode'? |
| 09:29 | <Hixie> | i don't want to change the foreign stuff today |
| 09:29 | <Hixie> | or at all, until the svg issue can be resolved |
| 09:36 | hsivonen | wonders how Google Chrome will do video codecs |
| 09:37 | <hendry> | really daft how http://www.google.com/googlebooks/chrome/# isn't addressable |
| 09:38 | <hsivonen> | no alt attribute there |
| 09:41 | <hsivonen> | hendry: It's really weird how Google is now publishing content in a form that is unfriendly towards search engines |
| 09:41 | <hsivonen> | hendry: and so far, it seems that Googlebot isn't getting any unfair magic advantage |
| 09:42 | <hsivonen> | the stuff just doesn't work with bot-based search |
| 09:42 | <Hixie> | google would never give googlebot an unfair advantage on one of its own sites |
| 09:42 | <Hixie> | in fact the search team goes out of its way to avoid doing that |
| 09:43 | <Hixie> | ok |
| 09:43 | <Hixie> | i need to solve the </script> problem for html now |
| 09:43 | <annevk> | incidentally, is that why Google uses crap markup? :D |
| 09:43 | <Hixie> | i guess i just say that a parser-inserted <script> element doesn't do magic when inserted? |
| 09:44 | <Hixie> | annevk: http://www.google.com/privacy/ |
| 09:44 | <Hixie> | annevk: we're learning. slowly. |
| 09:44 | <Hixie> | and then i can trigger the magic when i see </script> |
| 09:45 | <hsivonen> | I learned that Google Maps has the HTML5 doctype by downloading and parsing random stuff from dmoz |
| 09:45 | <Hixie> | having the doctype is one thing |
| 09:45 | <Hixie> | actually being compliant is what matters |
| 09:46 | <hsivonen> | yes, but I was still surprised to see Google do Ajaxy stuff in the standards mode |
| 09:46 | <Hixie> | true |
| 09:46 | <hsivonen> | given how recent that is in GWT |
| 09:48 | <annevk> | interesting |
| 09:49 | <Hixie> | actually i don't know if i told you this earlier, hsivonen, but someone in the zurich office set up an instance of your html5 validator internally |
| 09:50 | <Hixie> | to encourage html5 compliance |
| 09:50 | <hsivonen> | Hixie: cool |
| 09:50 | hsivonen | wonders if Jens Meiert from the WG is in the Zurich office |
| 09:50 | <hendry> | hsivonen: i hope too google will "provide leadership" with this video codec mess... |
| 09:50 | <Hixie> | hsivonen: yes; it was him who set up the validator in fact |
| 09:56 | <Philip`> | http://blogoscoped.com/files/google-chrome-screenshot.jpg - why does everyone want to get rid of menus nowadays :-( |
| 09:56 | <Philip`> | s/ :/? :/ |
| 09:58 | <hsivonen> | how do multiple processes share drawing surfaces? |
| 09:58 | <annevk> | how do multiple processes still allow interaction with each other... eg, Flash communication |
| 09:59 | <hsivonen> | I've been told that the reason why Gecko's threading is the way it is was that back in 1998, X11 wasn't robust enough to handle even one *thread* per one *window* |
| 09:59 | <annevk> | maybe it's not completely sandboxed as the comic leds you to believe |
| 10:00 | Philip` | guesses it involves IPC |
| 10:01 | Philip` | has no idea of anything more detailed, though |
| 10:01 | <hsivonen> | or perhaps they overlay the content area somehow and don't share actual drawing surfaces |
| 10:01 | <hsivonen> | after all, there are already out-of-process browser plugins on Linux |
| 10:01 | <roc> | hsivonen: I don't know who told you that, but I doubt it. |
| 10:02 | <roc> | Having child windows (tabs) be rendered by different processes is actually "easy" on Windows and X11 |
| 10:02 | <Philip`> | Opera already has an operapluginwrapper process on Linux, which seems to be related to Flash plugins in some way |
| 10:03 | <roc> | might get tricky if your UI has transparent overlays over the content |
| 10:03 | <roc> | there is really no solution to that on Mac that I know of, so I'm looking forward to seeing how Chrome does it |
| 10:03 | <roc> | so windowed plugins are in the same category, so OK for Windows and X11 |
| 10:03 | <hsivonen> | roc: why does Gecko run all the windows on one thread? |
| 10:03 | <roc> | but *windowless* plugins, ... that |
| 10:04 | <roc> | that's a big problem. You can push bitmaps through shared memory, but that's suboptimal |
| 10:04 | <annevk> | I believe Chrome is Windows only for now |
| 10:04 | annevk | doesn't remember where he read that |
| 10:05 | <roc> | the beta is Windows only, but there is supposed to be Mac and Linux versions on the way |
| 10:05 | <roc> | hsivonen: the #1 reason is that making all structures shared across windows be threadsafe is a real pain |
| 10:06 | <annevk> | roc, indeed, I can't read :/ |
| 10:06 | <hsivonen> | roc: ok |
| 10:06 | <roc> | the #2 reason is that the Firefox UI, and especially the extension system, assumes a single-threaded model for all content |
| 10:06 | <hsivonen> | roc: but that came a lot later |
| 10:07 | <roc> | but anyway you don't really want one thread per window. Chrome is doing the right thing, you want process separation |
| 10:08 | <roc> | unfortunately, doing HW accelerated rendering from separate processes into the same window, with plugins in the mix too, is going to be awfully tricky |
| 10:09 | <roc> | hsivonen: those issues are fixable though, and I think we'll tackle them, possibly once we've had time to look at Chrome and learn what we can from it. |
| 10:10 | <hsivonen> | it seems that supporting the Flash binary blog is getting more and more troublesome |
| 10:11 | <hsivonen> | <video> obsoletes the video-only plug-ins |
| 10:11 | <hsivonen> | and some newer plugins like Gears and the Garmin GPS stuff are effectively faceless |
| 10:11 | <roc> | Process separation isn't free, there are performance and complexity costs, especially memory costs. Quite possibly it didn't make sense in the old days with small memories and only a few windows open at a time. |
| 10:11 | <roc> | Gears is not an NPAPI plugin. |
| 10:11 | <hsivonen> | oh. I thought it was |
| 10:12 | <roc> | Process separation probably still doesn't make sense for mobile. |
| 10:13 | <roc> | it'll be a long time before we can get rid of Flash (or Java, or Silverlight). I'm more interested in working with the plugin vendors to get better APIs so there isn't as much friction. |
| 10:16 | <hsivonen> | isn't Java now an intranet problem? |
| 10:16 | <hsivonen> | I use the Web with Java disabled |
| 10:16 | <hsivonen> | and it doesn't Break the Web |
| 10:17 | <hsivonen> | I also don't have Silverlight and not having it doesn't Break the Web, either |
| 10:17 | <Philip`> | Lots of web sites teaching physics still use Java applets |
| 10:17 | <othermaciej> | Gears has to hook into browsers in ways that go beyond any public plugin API |
| 10:18 | <hsivonen> | Philip`: yeah, some university demo pages are really retro |
| 10:18 | <othermaciej> | roc: I asked one of our mobile guys about multiprocess browser and he looked at me funny |
| 10:18 | <roc> | It may not break the Web for you, but when Java breaks, people yell |
| 10:19 | <hsivonen> | Philip`: I did have to use Java applets for math and computer science demos while studying |
| 10:19 | <hsivonen> | Philip`: but all those would be better as Java Web Start apps or as JavaScript apps |
| 10:20 | <roc> | othermaciej: Perhaps the mobile people are still getting used to operating systems with address space separation |
| 10:20 | <hsivonen> | I still haven't seen an applet that shouldn't be rewritten in JS or put into a dedicated Java Web Start window |
| 10:20 | <othermaciej> | roc: iPhone has plenty of address space separation, just not plenty of memory |
| 10:20 | <othermaciej> | it wasn't made by classical style "mobile people" |
| 10:20 | <hsivonen> | Philip`: those were Netscape 4-era applets |
| 10:21 | <roc> | smart move |
| 10:22 | <Hixie> | a number of online games i play (e.g. scrabbulous) use java |
| 10:23 | <hsivonen> | Hixie: do they feel like they should be Web Start apps of GWT+canvas stuff instead? |
| 10:23 | <Hixie> | most could just be normal html web apps like gmail or google calendar |
| 10:26 | <hsivonen> | Hixie: btw, do the games you play use <object> or <applet>? |
| 10:27 | <Hixie> | good question |
| 10:30 | <Hixie> | scrabbulous uses both |
| 10:30 | <Hixie> | object for ie, applet for others |
| 10:31 | <othermaciej> | I think my only experiences of Java and Flash on the Web are games and ads |
| 10:31 | <Hixie> | toulouse uses just applet |
| 10:31 | <roc> | and video I presume |
| 10:31 | <othermaciej> | some Flash games seem like they would be hard to do with good performance in current gen browsers |
| 10:31 | <othermaciej> | and video, yeah |
| 10:31 | <othermaciej> | (hard to do on open standards Web tech that is) |
| 10:31 | <Hixie> | we're getting there |
| 10:31 | <Hixie> | slowly :-) |
| 10:32 | <roc> | quite a few sites use Flash as a sound or persistent storage API (including gmail) |
| 10:32 | <hsivonen> | Hixie: Sun still says <applet> works best cross-browser: http://java.sun.com/docs/books/tutorial/deployment/applet/deployindex.html |
| 10:32 | <Hixie> | hsivonen: the arguments against making <applet> conforming are more political than anything else |
| 10:33 | <hsivonen> | othermaciej: do you count branded product catalogs entangled in Flash as advertising? |
| 10:33 | <othermaciej> | hsivonen: I have not used any such lately |
| 10:34 | <othermaciej> | roc: I guess that's not visible to me, but that is kind of sad |
| 10:34 | <roc> | well, we're all doing our best to change that |
| 10:36 | <roc> | does using a Java telnet applet to play nethack count as a "game"? |
| 10:41 | <Hixie> | ok, checked in the revamp of <script> handling |
| 10:41 | <othermaciej> | roc: I am happy with where the open Web platform is going |
| 10:42 | <roc> | good. I'm scared |
| 10:42 | <othermaciej> | and I've always enjoyed collaborating on actual standards work with Mozilla folks |
| 10:42 | <othermaciej> | and I think Gecko and WebKit are going in gratifyingly similar directions strategically despite pretty much no explicit coordination |
| 10:43 | <othermaciej> | (this despite occasional personality clashes or the like w/ specific Mozilla folks) |
| 10:47 | <aaronlev> | hsivonen: what was the discussion about aria-setsize/aria-posinset? |
| 10:49 | <hsivonen> | aaronlev: are you referring to Ben Millard's feedback to the PFWG? |
| 10:50 | <hsivonen> | aaronlev: IIRC, there was (months ago) discussion about setsize belonging on the container instead of being repeated on each item |
| 10:50 | <hsivonen> | aaronlev: and instead of posinset, an analogue of the start attribute on <ol> would be less verbose for paged views |
| 10:52 | <aaronlev> | hsivonen: ok, i saw an ongoing discussion which was forwarded to me |
| 10:52 | <aaronlev> | problem is sort of related to how trees or tree grids are structured and whether the whole thing is loaded at once |
| 10:53 | <aaronlev> | sometimes, some of the structure is missing |
| 10:53 | <hsivonen> | aaronlev: isn't it weird to have to update setsize in many places? |
| 10:53 | <aaronlev> | yes it's weird |
| 10:53 | <aaronlev> | ajax trees are weird |
| 10:53 | <aaronlev> | the structure is half gone |
| 10:54 | <aaronlev> | only the server knows the structure |
| 10:54 | <aaronlev> | but i agree we should look for a way to make it better |
| 10:57 | <aaronlev> | hsivonen ^ |
| 10:58 | <aaronlev> | basically trees and treegrids are a case where there's often so much data on the server it's cruicial perf by only loading the visible lines, which could be anywhere in the middle |
| 10:59 | <hsivonen> | aaronlev: I guess I don't know enough about ajax trees to see how there wouldn't be a container element to hand setsize onto |
| 10:59 | <hsivonen> | s/hand/hang/ |
| 10:59 | <aaronlev> | hsivonen: we should look at them again, maybe we can enforce that |
| 10:59 | <aaronlev> | i think in our case we're just dealing with how authors do things in JS |
| 11:00 | <aaronlev> | not what an ideal HTML widget would do |
| 11:00 | <aaronlev> | so, HTML would probably have a cleaner solution |
| 11:00 | <aaronlev> | aria is sometimes just accepting of the fact that it's applied to stuff after it was already coded, that a11y was an afterthought |
| 11:01 | <aaronlev> | but i'm definitely willing to help look at ways of making that better |
| 11:03 | <Hixie> | search for "XSLT" in http://www.whatwg.org/specs/web-apps/current-work/#the-doctype and lower -- is it ok? |
| 11:03 | <hsivonen> | Hixie: if an non-parser-inserted script element runs as the result of insertion and it itself inserts another script element that runs immediately, should control return to the first-mentioned script afterwards or should it die? |
| 11:03 | <Hixie> | why would it die? |
| 11:04 | <hsivonen> | Hixie: it seems to die in the HTML5 Live DOM in Gecko |
| 11:04 | <Hixie> | really? that's odd |
| 11:04 | <hsivonen> | Hixie: I can't tell if it's my bug, Gecko bug or a platform feature |
| 11:04 | <Hixie> | testcase? |
| 11:08 | <hsivonen> | Hixie: hmm. might be my problem. I have to investigate more. |
| 11:08 | <Hixie> | k |
| 11:09 | <Hixie> | bed time for me |
| 11:09 | <Hixie> | nn |
| 11:09 | <hsivonen> | Hixie: <!DOCTYPE HTML PUBLIC "XSLT-generated"> WFM |
| 11:09 | <hsivonen> | nn |
| 11:10 | <annevk> | still seems silly to cater for XSLT, just output XML |
| 11:10 | <hsivonen> | annevk: IE |
| 11:11 | <hsivonen> | "just output XML" isn't realistic |
| 11:11 | <annevk> | IE supporting XML at some point is realistic though |
| 11:12 | <hsivonen> | XSLT users will need validation before that |
| 11:12 | <hsivonen> | *way* before |
| 11:12 | <Philip`> | If IE supports XHTML before HTML5 goes to Rec, we can just remove that catering-for-XSLT feature |
| 11:13 | <hsivonen> | Philip`: at which point XSLT people would be angry at me for their stuff no longer validating |
| 11:14 | <Philip`> | hsivonen: That's their fault for relying on a draft spec |
| 11:14 | <annevk> | therefore it seems like a bad thing to add it now |
| 11:14 | <annevk> | oh well |
| 11:14 | <hsivonen> | Philip`: that's not really a tenable position when <video> is a killer feature happening long before REC |
| 11:15 | <annevk> | <video> doesn't work with XSLT anyway |
| 11:15 | <hsivonen> | annevk: <video> does. <source> doesn't |
| 11:16 | <annevk> | <source> is sort of mandatory at this point |
| 11:16 | <hsivonen> | annevk: <video src> should be just fine with XSLT |
| 11:16 | <annevk> | I know |
| 11:16 | <hsivonen> | annevk: it isn't, if you are doing Ogg-only |
| 11:17 | <Philip`> | Hixie: You just added <code titel=""> |
| 11:17 | <hsivonen> | annevk: what's the point of an RF codec on the authoring side if everyone makes an encumbered alternative anyway? |
| 11:18 | <annevk> | hsivonen, dunno, we'll see how it goes I suppose |
| 11:19 | <Philip`> | Hixie: "The same character as in item 8" - s/8/9/ |
| 11:20 | <Philip`> | Hixie: Also, in "For the purposes of XSLT generators, which cannot output HTML markup without a DOCTYPE", maybe remove the comma because otherwise it's incorrectly suggesting that no XSLT generator is able to output HTML markup without a DOCTYPE |
| 11:21 | <roc> | hsivonen: you can fall back to .WAV! |
| 11:38 | <hsivonen> | argh. the new tokenizer tweaks pushed the loop 8 bytes over the magic hotspot limit again (with ecj) |
| 11:42 | Philip` | doesn't think that would count as a valid technical reason to change the spec |
| 11:42 | <hsivonen> | Philip`: I'm going to extract methods for errors |
| 11:42 | <Philip`> | (Isn't there a rarely-called chunk of code you could just split out into a separate method, to keep the main loop small?) |
| 11:42 | <hsivonen> | Philip`: so all the string formatting code won't count |
| 11:43 | <hsivonen> | Philip`: besides, I should do the null check for error handler *before* I format the error string |
| 11:44 | <hsivonen> | also, the error reporting code needs a much better localization story |
| 11:48 | <hsivonen> | Philip`: also, the spec change in this case was something I asked for |
| 11:48 | <hsivonen> | (and I could just not care about perf with ecj and only care about perf with javac) |
| 11:49 | <hsivonen> | ecj does silly stuff like goto n; n: goto m |
| 12:17 | <Lachy> | Hixie, why did you go with <!DOCTYPE html PUBLIC "XSLT-generated"> instead of an emtpy string? |
| 12:19 | <Lachy> | Hixie, why does the spec give all the Unicode names for the characters in the DOCTYPE, but then only say "The literal string "XSLT-generated""? |
| 12:20 | <hsivonen> | Hixie: why is XSLT-generated an ASCII-case-insensitive match when doctype quirkiness check is? |
| 12:20 | <Lachy> | oh, nevermind, I guess it's because the XSLT-generated is case sensitive, whereas the rest of it isn't. |
| 12:20 | <hsivonen> | *why isn't |
| 12:23 | <hsivonen> | never mind. I added a case-sensitive check to my string portability layer |
| 12:29 | <hendry> | http://lists.w3.org/Archives/Public/public-mwts/2008Sep/0000.html # any suggestions for a test suite to do with mobiles? |
| 12:30 | <hendry> | now the http://www.w3.org/2008/06/mobile-test/ "Web Compatibility Test for Mobile Browsers" is done ... |
| 12:31 | <hsivonen> | hendry: shouldn't mobiles be tested against the same suites as desktop browsers? |
| 12:32 | <hsivonen> | (except for CSS tests where the test assumes a wider viewport) |
| 12:32 | <hendry> | hsivonen: yes, they should be really |
| 12:33 | <hendry> | hsivonen: though i guess we're looking to highlight areas particularly lacking in mobiles |
| 12:34 | <hendry> | so there was earlier talk on "performance" testing, but i'm not so sure |
| 12:40 | <hsivonen> | I wonder what the license for http://www.croczilla.com/svg/samples/lion/lion.svg is |
| 12:48 | <Philip`> | hendry: Maybe mobiles are lacking in <canvas> support? :-) |
| 12:50 | <hsivonen> | is tiger.eps under the GPL when obtained from GNU Ghostscript? |
| 12:51 | <hendry> | Philip`: there is canvas test already #14 http://www.w3.org/2008/06/mobile-test/doc.html -- perhaps it can be improved? |
| 12:54 | <Philip`> | hendry: If the aim is to test that it's implemented entirely correctly, as opposed to simply being implemented at all, then I've got six hundred test cases for specific canvas features :-) |
| 12:56 | <wilhelm> | Philip`: The aim is to test whether various features are present, to determine whether a mobile browser is ready for the "one web", not just the "mobile web". The other six hundred tests for each feature is out of scope. (c: |
| 12:56 | <hendry> | well said wilhelm |
| 12:57 | <Philip`> | Okay - as long as it doesn't encourage mobile browser developers to just provide minimal stub implementations of all those features, that sounds reasonable :-) |
| 12:59 | <wilhelm> | It would be possible to cheat, of course. Opera unintentionally cheats on the inputmode test. But where possible, the document links to relevant test suites. |
| 13:01 | <hsivonen> | wilhelm: how do you cheat? do you implement it per WF2 instead of XHTML Basic 2nd ed? |
| 13:03 | <wilhelm> | We support the feature in the browser core, so it's present in the DOM. But there is no UI for it, so it can't actually be used. |
| 13:09 | <Lachy> | hmm, this headers thread is going round in circles again. We already know scope isn't well supported today. http://lists.w3.org/Archives/Public/public-html/2008Sep/0046.html |
| 13:09 | <hsivonen> | http://hsivonen.iki.fi/test/perf.html |
| 13:10 | <Lachy> | but it would be nice to get an experemental implementation of the smart headers algorithm in a browser which can be tested with an actual screen reader |
| 13:12 | <Philip`> | Lachy: Why would that be better than having an offline implementation, that can transform a table to just use @headers so that screen readers will get the same header associations as if they ran the more complicated algorithm? |
| 13:13 | <hsivonen> | Philip`: why offline? greasemonkey could work |
| 13:13 | <Philip`> | hsivonen: Ah, that's true, though it wouldn't help with IE-based screen readers |
| 13:13 | <Lachy> | Philip`, to prove that even with the costly and long upgrade cycle of assistive technology, updating the browser it uses to provide the support is a workable solution |
| 13:14 | <hsivonen> | hmm. it seems that the bottleneck with the Tiger in Gecko is DOM or painting, because TraceMonkey doesn't seem to make it faster |
| 13:15 | <Philip`> | Lachy: Ah, okay |
| 13:16 | <wilhelm> | hsivonen: Could you return the loading time on that page, for easy comparision between builds and browsers? |
| 13:19 | <hsivonen> | wilhelm: yeah. I'll improve it later. gotta go now |
| 13:56 | <hendry> | on the video topic; http://www.fsf.org/resources/formats/playogg and http://stephenfry.com/blog/?p=52 today |
| 13:59 | <hsivonen> | hendry: doesn't play on page in minefield on Mac |
| 14:03 | <hendry> | hsivonen: nightmare :/ |
| 14:06 | <hsivonen> | Lachy: note how the FSF is using CC-by-nd here but not NC |
| 14:07 | <Lachy> | hsivonen, note how you're using that appeal to authority again! :-) |
| 14:13 | <annevk> | is there a MacBook Air next to him? |
| 14:28 | <hsivonen> | Lachy: as far as authorities go, Stephen Fry seems like a better celebrity endorser than Jerry Seinfeld :-) |
| 14:29 | <Lachy> | hsivonen, I don't know who Stephen Fry is |
| 14:29 | <Philip`> | :-o |
| 14:32 | <gavin_> | seinfeld's much more popular here in the americas |
| 14:33 | <gavin_> | and in the australias too, I guess |
| 14:35 | <zcorpan> | Hixie: "The same character as in item 8" - s/8/9/ |
| 14:35 | <Philip`> | I've heard of a TV show called "Seinfeld" and I can assume Jerry Seinfeld is related to that, but other than that I have no idea who he is |
| 14:35 | <Philip`> | zcorpan: http://krijnhoetmer.nl/irc-logs/whatwg/20080902#l-443 :-) |
| 14:36 | <hsivonen> | Philip`: that's the one. he'll promote Vista. |
| 14:36 | <hsivonen> | and the show Seinfeld has been gone for years but Fry is still shooting new stuff |
| 14:36 | <zcorpan> | Philip`: hah |
| 14:37 | <annevk> | the show Seinfeld is pretty awesome though |
| 14:37 | <hsivonen> | annevk: sure |
| 14:39 | <Lachy> | Philip`, what country are you from? How could you not know who Jerry Seinfeld i? |
| 14:39 | <Lachy> | s/i/is/ |
| 14:41 | <Philip`> | Lachy: England, where I assume it hasn't been shown much or if it has then I don't remember ever seeing it |
| 14:41 | <Philip`> | How could you not know who Stephen Fry is? :-p |
| 14:42 | <gavin_> | I think I recall seeing something with stephen fry in it |
| 14:42 | <gavin_> | or maybe I just found out about him after looking up hugh laurie |
| 14:43 | <Philip`> | Blackadder? |
| 14:43 | <Lachy> | Philip`, Australians get a lot more American TV shows than English shows |
| 14:44 | <Lachy> | English TV is largely limited to being broadcast on the ABC, which is far less popular than the 3 commercial TV networks |
| 14:44 | <gavin_> | never heard of "blackadder" |
| 14:44 | <Philip`> | He narrated the UK editions of the Harry Potter audiobooks too, but I don't know how many other countries use the UK rather than US version (or do they have their own new version?) |
| 14:45 | <hsivonen> | Philip`: was Stephen Fry in Blackadder? Hugh Laurie was |
| 14:45 | <Philip`> | hsivonen: Wikipedia says he was, so it must be true |
| 14:46 | <Philip`> | (Also it has a photo of him in it, which is quite convincing) |
| 14:46 | <hendry> | Fry is prolific. His books are great too. |
| 14:46 | <hsivonen> | Philip`: ah right. fourth season |
| 14:47 | hsivonen | is surprised to learn that English TV isn't big in Australia |
| 14:48 | gavin_ | too |
| 14:48 | Philip` | can only actually remember watching one episode of Blackadder, and that was in a history lesson at school |
| 14:49 | <hsivonen> | Philip`: you had good history material at school :-) |
| 14:49 | <Philip`> | (but at least I've seen all of QI) |
| 14:50 | <Philip`> | We watched The Matrix in a maths lesson too |
| 14:51 | <hsivonen> | hmm. that sounds questionable |
| 15:00 | <Lachy> | I'd heard of Blackadder, and I'm pretty sure it was aired in Australia, but I never watched it, probably because I was too young to be interested in it when it came out |
| 15:00 | <Lachy> | "The Black Adder was the first series of Blackadder and was written by Richard Curtis and Rowan Atkinson, and produced by John Lloyd. The series was originally aired on BBC 2 from 1983-06-15 to 20 July, 1983, and was a joint production with the Australian Seven Network." |
| 15:01 | <Lachy> | I didnt' know it was jointly produced by channel 7 |
| 15:26 | <zcorpan> | standardssuck.org needs a new entry |
| 15:26 | <zcorpan> | 4 aug |
| 15:26 | <zcorpan> | almost a month |
| 15:27 | <Lachy> | zcorpan, I know. I spoke to marcos about it last night |
| 15:28 | <zcorpan> | Lachy: ok |
| 15:28 | <Lachy> | if you have any suggestions for who to interview and what topic, let me know |
| 15:29 | <zcorpan> | how about svg? |
| 15:30 | <Lachy> | yeah, maybe I could interview Erik D about it |
| 15:30 | <annevk> | isn't zcorpan closer to ed? |
| 15:31 | <zcorpan> | i am |
| 15:31 | <Lachy> | oh, is he in Sweden? |
| 15:31 | <annevk> | yes |
| 15:31 | <Lachy> | I thought he was in Oslo |
| 15:31 | <annevk> | no |
| 15:31 | <Lachy> | zcorpan, are you able to do it? |
| 15:31 | <annevk> | (the reason for the break was simply vacation) |
| 15:31 | <annevk> | (i tell myself) |
| 15:31 | <zcorpan> | Lachy: dunno, don't have a camera for one |
| 15:31 | <Lachy> | ok |
| 15:31 | <annevk> | zcorpan, ed has a MacBook, should work |
| 15:31 | <zcorpan> | annevk: ah |
| 15:32 | <zcorpan> | i could ask if he's up for it |
| 15:32 | <Lachy> | yeah, as long as he as either QuickTime Pro or iMovie |
| 15:33 | <Lachy> | we should be able to get several interviews done when I go to Web Directions and then there's TPAC soon |
| 15:40 | Dashiva | wonders if any signficant amount of AT software will support @scope by the time of HTML6... |
| 15:42 | <Lachy> | Dashiva, AIUI, it depends on what the browser reports through the accessibility API, so if the browser supports it, it should be able to report the necessary info. But I could be wrong. |
| 15:47 | <Dashiva> | Then why are people saying AT don't support it, rather than browsers? |
| 15:48 | <Dashiva> | It's not like scope is used much outside of accessibility APIs |
| 15:48 | <Lachy> | I don't know. I could be completely wrong |
| 15:50 | <Lachy> | but that seems to be the way ARIA works, cause the browser sees the attributes and then reports to the AT what something is based on that. Apparently, ATs don't access the DOM directly, they do everything through teh accessibility API |
| 16:04 | <zcorpan> | Lachy: they do some things by accessing the dom |
| 16:04 | <zcorpan> | aiui |
| 16:04 | <zcorpan> | but they already have code that acts on the accessibility api |
| 16:05 | <zcorpan> | so it's less work for AT vendors if the browser does the mapping |
| 16:12 | <takkaria> | oh, damn, there's another character to turn into U+FFFD at the inputstream stage |
| 16:13 | <Philip`> | You could treat it as a fatal error instead |
| 16:14 | <takkaria> | that's got the same implementation cost, more-or-less |
| 16:15 | <zcorpan> | perhaps one day we'll report errors for non-Name tags |
| 16:15 | <takkaria> | in Hubbub, the inputstream just passes a UTF-8 stream of characters with no filtering, and U+0000 is handled in the same way as any other character would be, it just generally inserts U+FFFDs around the place |
| 16:16 | <takkaria> | though all I need to do is change "c == '\0'" to "c == '\0' || c == '\x0b'" for the most part |
| 16:23 | <Lachy> | annevk, would talking about Google Chrome in a standardssuck episode be on topic, even though it's a browser, not a standard? |
| 16:24 | <Lachy> | hmm, maybe if we focuss on their standards support. I wonder if google gears has implemented the HTML5 api yet |
| 16:27 | <Philip`> | Wait a few hours, then it should be clearer how much they diverge from WebKit |
| 16:33 | myakura_ | wondering if they use JavaScriptCore for V8 |
| 16:43 | <Philip`> | myakura_: The API, or the implementation? |
| 16:48 | <myakura_> | Philip`: meant the latter, but if to know about both is fine :) |
| 19:00 | <billyjack> | waiting for Chrome videocast announcement |
| 19:00 | <billyjack> | not happened yet, right? |
| 19:00 | <billyjack> | KevinMarks: ? |
| 19:01 | <MikeSmith> | "Another one bites the dust" finishes... onto to some "yeah, you got satin shoes..." |
| 19:02 | MikeSmith | wonders whose iPod the background tunes are coming from |
| 19:04 | <MikeSmith> | is it not now already 11am Pacific? |
| 19:05 | <gavin_> | it is |
| 19:05 | <KevinMarks> | the videocast url is up, nothing showing on it yet |
| 19:06 | <MikeSmith> | KevinMarks: looks like it's running now |
| 19:07 | <MikeSmith> | Sundar |
| 19:07 | <hober> | Where did they post the videocast url? |
| 19:07 | <KevinMarks> | hm |
| 19:07 | <KevinMarks> | what url you got, MikeSmith? |
| 19:07 | <KevinMarks> | I'm getting "Whole lotta love" on the one I have |
| 19:08 | <KevinMarks> | http://google.client.shareholder.com/Visitors/event/build2/MediaPresentation.cfm?MediaID=33101&Player=1# |
| 19:08 | <MikeSmith> | KevinMarks: http://google.client.shareholder.com/Visitors/event/build2/MediaPresentation.cfm?MediaID=33101&Player=1 |
| 19:08 | <MikeSmith> | or http://google.client.shareholder.com/Visitors/event/build2/MediaPresentation.cfm?MediaID=33101&Player=2 |
| 19:09 | <MikeSmith> | I was getting Whole lotta love too before they went live |
| 19:09 | <MikeSmith> | KevinMarks: force reload? |
| 19:10 | <KevinMarks> | let me try it in firefox ;) |
| 19:11 | <KevinMarks> | aha, that works |
| 19:12 | <KevinMarks> | I bet safari was caching the old mov |
| 19:12 | <MikeSmith> | KevinMarks: :) |
| 19:13 | <MikeSmith> | Sundar now asking the question, Why did we choose WebKit? |
| 19:14 | <Philip`> | Because building on Trident would be a really bad idea? |
| 19:15 | <jcranmer> | how about Tasman? |
| 19:16 | <MikeSmith> | he's talking about V8 now |
| 19:17 | <MikeSmith> | says Lars will give more details in the videocast later |
| 19:17 | <Philip`> | Is it "V8" as in "V8 engine", or is it stupid txt-speak for "vate"? |
| 19:17 | <MikeSmith> | (me hopes somebody asks whether it's faster than Squirrelfish... or least if it has a better logo (which I highly doubt)) |
| 19:18 | <gsnedders> | Is it faster than SquirrelFish? |
| 19:18 | <Philip`> | Is it faster than TraceMonkey? |
| 19:18 | <Dashiva> | Philip`: The image was an engine, so probably the former |
| 19:18 | <Philip`> | Is it faster than JScript.NET? |
| 19:18 | <gsnedders> | Is it faster than C? |
| 19:18 | <Philip`> | Is it faster than God? |
| 19:19 | <Dashiva> | Definitely |
| 19:19 | <MikeSmith> | "Chromium" project name |
| 19:19 | <Dashiva> | I mean, when was the last god did a release? |
| 19:19 | <Dashiva> | *last time |
| 19:19 | <MikeSmith> | Ben Goodger takes the floor |
| 19:21 | <MikeSmith> | who's this Brian guy? |
| 19:21 | <Philip`> | Brian Blessed? |
| 19:21 | <gsnedders> | The Messiah? |
| 19:21 | <MikeSmith> | tab-dragging.. I suspect most people are thinking ho hum |
| 19:21 | <gsnedders> | MikeSmith: I am. |
| 19:22 | <gsnedders> | wow! Resizing tabs! Groundbreaking! |
| 19:22 | Philip` | has a grand total of one browser window, so there's nowhere to drag tabs to |
| 19:22 | <gsnedders> | Philip`: No, just reordering them within a window |
| 19:23 | <Philip`> | gsnedders: Reordering tabs within a window is quite substantially unexciting |
| 19:23 | <gsnedders> | Philip`: Thus the "ho hum" |
| 19:23 | <gsnedders> | an ombibox! |
| 19:24 | <Philip`> | Is that the bit that's like every other browser's latest version's address box? |
| 19:24 | <gsnedders> | No |
| 19:27 | <Dashiva> | The implicit multisearch is new, the rest seems done before |
| 19:27 | <Philip`> | Multisearch? |
| 19:29 | MikeSmith | wanna hear about the process manager and V8 |
| 19:29 | <gsnedders> | hah! |
| 19:29 | <gsnedders> | e.g., pr0n. |
| 19:30 | <gsnedders> | ("sometimes you visit sites you wouldn't want other people who use the computer know you visit") |
| 19:30 | <MikeSmith> | "Hide This from My Girlfriend/Wife" feature |
| 19:32 | <virtuelv> | impressive how many features borrowed from others are always "innovative" |
| 19:33 | <gsnedders> | features are always innovative in any software |
| 19:33 | <MikeSmith> | all of this stuff was in Xanadu 50 years ago |
| 19:34 | <MikeSmith> | and/or built into GNU Emacs by RMS personally |
| 19:34 | <MikeSmith> | and predicted by Giordano Bruno in the 16th century |
| 19:34 | <gsnedders> | Good mic |
| 19:34 | <MikeSmith> | Darin! |
| 19:35 | <Dashiva> | and had a prototype designed by Leonardo himself |
| 19:35 | <MikeSmith> | Darin is shorter and younger than I imagined |
| 19:36 | <gsnedders> | Who's Darin? |
| 19:36 | <gavin_> | darin fisher, I'm guessing |
| 19:36 | <gavin_> | he used to work on Mozilla |
| 19:36 | <gsnedders> | I recognize the name, vaguely |
| 19:37 | <MikeSmith> | yeah, Darin Fisher |
| 19:37 | <MikeSmith> | gsnedders: he wrote the Gospel of Luke |
| 19:37 | <MikeSmith> | in the bible |
| 19:37 | <gsnedders> | MikeSmith: huh? |
| 19:37 | <MikeSmith> | that's probably where you remember him from |
| 19:38 | <MikeSmith> | he was one of the original disciple of Jesus |
| 19:38 | <gsnedders> | … |
| 19:38 | <MikeSmith> | virtuelv: Darin metion malware |
| 19:38 | <MikeSmith> | malware++ |
| 19:39 | <MikeSmith> | Task Manager screen shot |
| 19:39 | <Dashiva> | very nice++++++ |
| 19:40 | MikeSmith | feels full bottle of Suntory Toma Toma beginning to take effect |
| 19:40 | <MikeSmith> | "misbehaving" |
| 19:42 | <MikeSmith> | wow |
| 19:42 | <MikeSmith> | killing plugins |
| 19:42 | <MikeSmith> | cool |
| 19:42 | <gsnedders> | How long will it go on for? |
| 19:43 | <MikeSmith> | kill kill |
| 19:43 | <MikeSmith> | gsnedders: days |
| 19:43 | <gsnedders> | :D |
| 19:43 | <MikeSmith> | tag team |
| 19:43 | <virtuelv> | MikeSmith: I'm not sure I got that |
| 19:44 | <MikeSmith> | virtuelv: me either |
| 19:45 | <Dashiva> | Oh, the accentity |
| 19:45 | <Philip`> | http://www.google.com/chrome |
| 19:45 | <MikeSmith> | dudes named Lars are usually pretty good at writing browser code |
| 19:45 | <MikeSmith> | in my experience at least |
| 19:45 | <gsnedders> | oh sux. |
| 19:45 | <Philip`> | now exists, and tells me Google hates Linux and won't give me any software |
| 19:45 | <virtuelv> | "Google Chrome for Linux is in development and a team of engineers is working hard to bring it to you as soon as possible." |
| 19:46 | <virtuelv> | *sigh* |
| 19:46 | <MikeSmith> | if you want to write a kick-ass browser, hire somebody named Lars |
| 19:46 | <virtuelv> | why not just give me the windows download link |
| 19:46 | <gsnedders> | "What should I say?" |
| 19:46 | <Dashiva> | MikeSmith: Wouldn't it be the other way around? All the good Larses are already hired, so you should avoid the rest? :) |
| 19:46 | <Philip`> | virtuelv: Try http://www.google.com/chrome/eula.html |
| 19:46 | <gsnedders> | MikeSmith: I want a job! |
| 19:46 | <MikeSmith> | the second-tier Larses |
| 19:47 | <gsnedders> | LarsYngve: :P |
| 19:47 | <Dashiva> | Do most javascript engines really interpret? |
| 19:47 | <virtuelv> | gsnedders: you'd have to know those two people |
| 19:47 | <virtuelv> | hm, |
| 19:48 | <Dashiva> | I thought they all did at least some compiling |
| 19:48 | <virtuelv> | installer crashes under wine |
| 19:48 | <gsnedders> | virtuelv: Which two? |
| 19:48 | <Philip`> | Dashiva: If by "interpret" they mean "execute bytecode instead of machine code", then yes |
| 19:48 | <virtuelv> | Two guys from Opera |
| 19:48 | <virtuelv> | (well, one of them is with Adobe these days) |
| 19:48 | gsnedders | can't quite remember who Lars is |
| 19:49 | <Philip`> | Wow, I've always wanted to download an installer and then execute it and then have it start downloading more stuff itself, without even telling me how much it's downloading or how fast it's going |
| 19:50 | <Philip`> | Clearly it'd be silly to just have a single file you download and install and that's it |
| 19:51 | <Philip`> | "Type to search or enter a URL to navigate - everything just works." - what if I want to search for a URL? |
| 19:51 | <Dashiva> | Hmm... V8 is BSD, so even IE could include it? |
| 19:52 | <roc> | Dashiva: all browser JS implementations must compile JS source code to something else. Most of them compile it to an AST or bytecode |
| 19:52 | <Philip`> | Google Chrome on Acid3: 79/100 |
| 19:53 | <virtuelv> | that's webkit -9 months, or so |
| 19:53 | <LarsYngve> | "preserve user choice"++ |
| 19:53 | <roc> | the big deal about V8 is apparently that it compiles all the way to machine code (but Tracemonkey also does that) |
| 19:53 | <gsnedders> | I'd guess that's mainly down to the limits of V8 |
| 19:53 | <LarsYngve> | where is Kirkland? |
| 19:53 | <Philip`> | Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US) AppleWebKit/525.13 (KHTML, like Gecko) Chrome/0.2.149.27 Safari/525.13 |
| 19:53 | <roc> | Kirkland's near Redmond |
| 19:54 | <gsnedders> | LarsYngve: Literally it means Church-land in Scots :P |
| 19:54 | <LarsYngve> | whoah, Larry at the mic |
| 19:54 | <gsnedders> | http://en.wikipedia.org/wiki/Kirkland — that lists loads of places |
| 19:55 | <roc> | if they're talking about Google employees, they're in the Seattle Kirkland |
| 19:55 | <roc> | that's where Google set up an office to catch escaping Microsoft employees |
| 19:55 | <Philip`> | Hmph, found a bug in Canvex: if the framerate exceeds 100, the FPS counter breaks :-( |
| 19:55 | <Dashiva> | Was that with chrome? |
| 19:55 | <LarsYngve> | Kirkland should be renamed Kirk Kirkwood |
| 19:56 | <Philip`> | Dashiva: Yes |
| 19:56 | <Philip`> | Dashiva: (Safari might be the same; I don't remember testing it recently) |
| 19:56 | <Philip`> | Chrome doesn't have canvas.getImageData :-( |
| 19:57 | <gsnedders> | Philip`: dhyatt said in #webkit he thinks it's Saf3.1 WebKit |
| 19:57 | <Dashiva> | I wonder if they're using apple's canvas or they made their own |
| 19:57 | <LarsYngve> | or Curt Kirkwood |
| 19:58 | <virtuelv> | anyone tried benchmarking yet? |
| 19:58 | <gsnedders> | #chronium |
| 19:58 | <Philip`> | Dashiva: It doesn't seem to be using Apple's Windows CG port, so I guess it's doing something a bit different to work on Windows |
| 19:59 | <gsnedders> | Philip`: Apparently it is Saf3.1 WebKit, though |
| 19:59 | <CurtKirkwood> | mobile strategy |
| 20:00 | <CurtKirkwood> | "V8 has been ported to ARM and Intel" |
| 20:01 | <CurtKirkwood> | mention of W3C |
| 20:02 | <Philip`> | gsnedders: But it's a Windows port of it, and it's not Apple's port |
| 20:02 | <gsnedders> | Philip`: They also said they hadn't changed much from WebKit |
| 20:04 | <Philip`> | http://dl.google.com/chrome/plugins/plugins.xml - Silverlight's not in that list :-( |
| 20:04 | <virtuelv> | fails acid2? |
| 20:04 | <gsnedders> | extensions… |
| 20:04 | <Philip`> | virtuelv: Passes for me |
| 20:04 | <gsnedders> | Plan to use it |
| 20:04 | <gsnedders> | s/use it/have them/ |
| 20:04 | <virtuelv> | Philip`: a coworker saw red over the eyes |
| 20:05 | <Philip`> | virtuelv: Passes for me :-) |
| 20:05 | <Philip`> | (at least after waiting long enough for all the images to load) |
| 20:07 | <Philip`> | There's an Inspector, that shows you the DOM tree and styles and stuff, and the times and sizes of downloaded resources - is that new or is that from WebKit? |
| 20:07 | <gsnedders> | Philip`: WebKit |
| 20:08 | <Philip`> | The tab-close button is too small |
| 20:08 | <mal> | inspector is webkit |
| 20:08 | <Philip`> | Hmm, the Task Manager has a "Stats for nerds" link |
| 20:48 | <Lachy> | hey, looks like Google forgot to make the videos about google chrome public http://www.google.com/chrome/intl/en/features.html |
| 20:48 | <Lachy> | youtube is blocking access |
| 20:54 | <aaronlev> | looks like bugzilla isn't being used much to track HTML 5 issues |
| 20:54 | <Hixie> | it's used |
| 20:54 | <aaronlev> | i only see 15 issues open |
| 20:54 | <Hixie> | yeah most of the issues are in e-mail at http://whatwg.org/issues/ |
| 20:54 | <aaronlev> | ok |
| 20:55 | <Hixie> | and because i tend to close the bugzilla ones quicker than e-mail |
| 20:55 | <aaronlev> | how is bugzilla working out |
| 20:55 | <Hixie> | pretty good for me |
| 20:55 | <Hixie> | but then i knew it already |
| 20:55 | <aaronlev> | right |
| 20:56 | <aaronlev> | i want to have ARIA implementor guidelines use that |
| 20:56 | <Hixie> | cool |
| 20:57 | <aaronlev> | it's not the ARIA spec -- it's the doc i started to describe what user agents need to do |
| 20:57 | <aaronlev> | if exposing to platform APIs |
| 20:57 | <Hixie> | sounds like a spec to me :-P |
| 20:57 | <Hixie> | what's the ARIA spec if not the document that says what uas should do? |
| 20:57 | <aaronlev> | well, it deals with a lot of platform issues |
| 20:58 | <aaronlev> | eventually they will fold in parts of it to the ARIA spec |
| 20:58 | <aaronlev> | but for now it's a smaller group of just browser developers working on it so we can go faster |
| 20:58 | <aaronlev> | it's something we should use for HTML |
| 20:58 | <Hixie> | cool |
| 20:58 | <aaronlev> | e.g. how to expose HTML via MSAA |
| 20:58 | <Hixie> | yeah, sounds useful. keep me in the loop |
| 20:58 | <aaronlev> | sure |
| 20:59 | <aaronlev> | it's a wiki, you can subscribe to see changes |
| 20:59 | <aaronlev> | http://developer.mozilla.org/en/docs/ARIA_User_Agent_Implementors_Guide |
| 21:00 | <Hixie> | i get far too much wiki notifications already :-/ |
| 21:00 | <aaronlev> | :) |
| 21:18 | <Lachy> | I'm giving up on that video accessibility thread. I don't know how else to explain that there's no need to provide alt text for a poster frame, when what is really needed is a brief summary of the video content or at least an indication of its topic |
| 21:19 | <hober> | (which should be in the markup surrounding <video> anyway...) |
| 21:19 | <Lachy> | yeah, either that or its title attribute |
| 21:20 | <Lachy> | The way Leif keeps going, he's making it seem as if the poster frame and the video represent 2 different things, which is just silly |
| 21:24 | <Hixie> | well, chrome is causing acidtests.org (and thus whatwg.org) to get hammered |
| 21:24 | <Hixie> | but at least this time the server is staying up! |
| 21:24 | <Hixie> | it just isn't responding to http traffic |
| 21:24 | <gsnedders> | hah. |
| 21:24 | <Dashiva> | Silly Hixie, you of all people should've seen this coming :) |
| 21:24 | <Hixie> | i did |
| 21:25 | <Hixie> | when firefox came out the server crashed and burnt |
| 21:25 | <Hixie> | this time the server is happily up |
| 21:25 | <Hixie> | just not responding to http traffic because apache's out of ports |
| 21:25 | <Hixie> | which is fine by me |
| 21:25 | <gsnedders> | This is all putting me off the schoolwork I should be doing |
| 21:25 | <Hixie> | since it means i can still check e-mail, work on html5, etc |
| 21:29 | <Philip`> | All browsers should come with a pre-cached copy of the Acid tests, to avoid the pointless server load when ten thousand people run precisely the same test case against precisely the same browser code |
| 21:29 | <Hixie> | heh |
| 21:33 | Philip` | thinks it's peculiar for Hixie to argue that changing implementations is easier than changing specifications |
| 21:34 | <gsnedders> | Philip`: Changing specs are a pain in the ass. I agree. |
| 21:35 | <Lachy> | I tested google chrome with my local copy of acid3, and only scored 77 |
| 21:36 | <gsnedders> | Saf3.1 has some odd bugs that make it inconsistent |
| 21:37 | <Lachy> | Safari 3 is scoring 75 for me and safari 4 developer preview is scoring 100% |
| 21:37 | <Lachy> | Official Build 1583Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US) AppleWebKit/525.13 (KHTML, like Gecko) Chrome/0.2.149.27 Safari/525.13 |
| 21:39 | <Lachy> | Safari 3: Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10_5_4; en-au) AppleWebKit/525.18 (KHTML, like Gecko) Version/3.1.2 Safari/525.20.1 |
| 21:39 | <Lachy> | so the webkit in Google Chrome is slightly older than that |
| 21:40 | <Hixie> | Philip`: for an end user, it is |
| 21:40 | <Hixie> | Philip`: for an end programmer, i should say |
| 21:40 | hallvors | admits he just loaded Acid2 in Chrome .. |
| 21:41 | <gsnedders> | Lachy: 525.13 seems to be the Windows build |
| 21:41 | <Hixie> | Lachy: yeah chrome forked off webkit a few months ago |
| 21:41 | <Hixie> | Lachy: they'll sync again soon |
| 21:42 | <gsnedders> | eseidel was saying he was working on syncing it up |
| 21:42 | <Hixie> | yeah |
| 21:42 | <eseidel> | *long* ago |
| 21:42 | <eseidel> | our webkit is like from january |
| 21:42 | <eseidel> | or whenever 3.1.1 forked |
| 21:42 | <Lachy> | Hixie, since you work at google, have you seen any internal builds before now? |
| 21:42 | <Hixie> | yes |
| 21:42 | <eseidel> | Hixie sorrta walls himself off in some other building |
| 21:43 | <eseidel> | sadly, we never really see him :( (those of us who work on actual browser stuff) |
| 21:43 | <gsnedders> | A big wall to stop him shouting at you? |
| 21:43 | <Hixie> | i'd have played with it more if it was a mac browser :-P |
| 21:43 | <eseidel> | probably |
| 21:43 | <Hixie> | running it in a VM is a pain in the ass |
| 21:43 | <Lachy> | eseidel, do you work at google too? |
| 21:43 | <eseidel> | yes |
| 21:43 | <Hixie> | eseidel is google's webkit team |
| 21:44 | <eseidel> | ha |
| 21:44 | <Lachy> | ah, ok |
| 21:44 | <eseidel> | in many ways true :) |
| 21:44 | Hixie | installs chrome into his non-secret VM and deletes his secret VM |
| 21:45 | <gsnedders> | Hixie: So you have no more über-c00l non-existing software? |
| 21:45 | <Lachy> | is hixie.ch and whatwg.org all on the same server as acid3.acidtests.org ? |
| 21:45 | <Hixie> | Lachy: yes |
| 21:45 | <Hixie> | it's all on hixie.dreamhost.com |
| 21:46 | <Hixie> | man i wish chrome had smooth scrolling |
| 21:46 | <Lachy> | you need more servers |
| 21:46 | gsnedders | chants: Hixie! Hixie! Hixie!… |
| 21:46 | <Hixie> | Lachy: right now i don't pay for them |
| 21:46 | <Hixie> | Lachy: so if you can find me another free server... :-) |
| 21:46 | <gsnedders> | Hixie: Doesn't your employee have a server or two? |
| 21:47 | <gsnedders> | *employer, even |
| 21:47 | <gsnedders> | I am _not_ your slave! |
| 21:47 | <Lachy> | how do you get free hosting from dreamhost? |
| 21:47 | <Hixie> | gsnedders: i doubt i'll work for google my whole life |
| 21:48 | <Hixie> | Lachy: the referral bonuses i get balance out the costs |
| 21:48 | <Philip`> | Hixie: Dreamhost likely won't stay in business for your whole life, so you'll have to migrate your server anyway |
| 21:48 | <Lachy> | oh, nice |
| 21:49 | <Lachy> | I have an effectively free host with asmallorange.com |
| 21:49 | <gsnedders> | Hixie: Stop spamming us, please! I've heard you try and get us sign up often enough! |
| 21:49 | <Lachy> | but it has montly bandwidth and storage limiatations |
| 21:50 | <Hixie> | Philip`: dreamhost has so far outlasted one of my employment terms |
| 21:50 | <Hixie> | Philip`: (i.e. i'm glad i didn't host acid2 or whatwg on opera servers) |
| 21:51 | <Lachy> | Hixie, why would that be a problem? I'm pretty sure Opera is still around |
| 21:51 | <Lachy> | Would they have revoked your right to host it there after you left? |
| 21:52 | <Hixie> | i doubt it would have been as simple as doing nothing |
| 21:53 | <Lachy> | OMG! even though you gave Julian what (s)he wanted with the XSLT-generated doctype, he's still not happy |
| 21:53 | <hober> | yeah, just responded to that (to www-archive; didn't want to waste list time with it) |
| 21:55 | <Lachy> | FWIW, I really hope Julian takes the issue to the XSLT WG and gets them to provide a way to output <!DOCTYPE html>, and then hopefully we can drop this XSLT-generated nonsense before well before HTML5 is done |
| 21:56 | <hober> | agreed |
| 22:02 | <Lachy> | I really don't like how google chrome copied IE's address bar feature that renders the domain in black and path in grey |
| 22:02 | <Lachy> | it makes reading it harder |
| 22:02 | <Lachy> | I can't find a way to turn it off either |
| 22:05 | <Lachy> | Hixie, Requiem 1.7.4 has been released |
| 22:05 | <Hixie> | what does it change? |
| 22:06 | <Lachy> | not sure yet. I'll find it and see |
| 22:06 | <Hixie> | i won't bother upgrading unless it stops working |
| 22:07 | <Lachy> | http://thepiratebay.org/torrent/4374326/Requiem_1.7.4_iTunes_DRM_Remover |
| 22:09 | <Lachy> | yeah, I guess, since you don't have the i18n problem, it's probably not worth it yet |
| 22:10 | <Lachy> | but I'll get it anyway cause I like to have the latest |
| 22:11 | <Hixie> | i'm done dealing with julian |
| 22:11 | <Hixie> | i try to help him and he just acts like an ass |
| 22:11 | <Hixie> | does anybody other than julian actually care about this XSLT thing? |
| 22:12 | <Hixie> | i guess jirka does |
| 22:12 | <Hixie> | he was nice about it |
| 22:12 | <Hixie> | so i'll leave it in |
| 22:13 | <Dashiva> | Didn't it turn out Julian wasn't using XSLT, but something else? |
| 22:19 | <Lachy> | yeah, Julian using using javax.xml.transform |
| 22:19 | <Lachy> | don't ask me why he's using an XML library to serialise HTML though |
| 22:20 | <Dashiva> | If only Russell was here, he'd tell us he was using a teapot to seralize HTML |
| 22:30 | <annevk> | making everyone upgrade to a non-IE browser or getting IE fixed is the real solution for these people |
| 22:30 | <annevk> | they don't like HTML anyway |
| 22:32 | <Dashiva> | Speaking of XHTML (kinda): Was there ever an official ruling on whether appendix C applies to 1.1? |
| 22:34 | <Hixie> | didn't the xhtml2 group come up with some new xhtml1.x that had a new appendix c or something? |
| 22:35 | <webben> | Dashiva: isn't that what the new media types draft is partly about? |
| 22:35 | <annevk> | Dashiva, they're working on that, yes |
| 22:35 | <webben> | providing guidance on how to serve xhtml 1.1 as text/html? |
| 22:35 | <Dashiva> | I know they're working on stuff relating to it, I'm just wondering about the current 1.1 |
| 22:35 | <annevk> | Dashiva, the second edition of 1.1 would say that |
| 22:37 | Hixie | puts on his Firefox t-shirt to emphasise his vendor-neutralness |
| 22:38 | <annevk> | you sort of lost that position now |
| 22:38 | <annevk> | strike "sort of" |
| 22:38 | <gsnedders> | Hixie: You need to where five shirts at once, really |
| 22:38 | <annevk> | hah, as if we'd be fooled by free t-shirts! |
| 22:39 | <Hixie> | annevk: well, html5 started while i was at opera, so i'm sure i'll live |
| 22:39 | <othermaciej> | so your best bet is to hope Chrome remains under 1% market share |
| 22:39 | <Hixie> | hah |
| 22:40 | <Hixie> | anyway. off to work i go. |
| 22:40 | <Hixie> | bbl |
| 22:40 | <Dashiva> | Hmm... the latest second editon draft says SHOULD text/html or application/xhtml+xml, but refers to XMLMIME which says SHOULD NOT on text/html for 1.1 |
| 22:40 | <annevk> | http://code.google.com/p/chromium/ has a huge team |
| 22:40 | <annevk> | and suspiciously there's no source code link |
| 22:41 | <svl> | src lives here: http://src.chromium.org/viewvc/ |
| 22:41 | <svl> | or at least the browser of - there's a tarball as well, iirc |
| 22:42 | <annevk> | so googlechrome.* are not Google owned after all, hmm |
| 22:48 | <annevk> | heh, "Rietveld", apparently Guido van Rossum names projects after Dutch artists |
| 22:49 | <Dashiva> | "XHTML documents served as 'text/html' will not be processed as XML [XML10], e.g. well-formedness errors may not be detected by user agents." |
| 22:49 | <Dashiva> | They say that like it's a bad thing |
| 22:50 | <csarven> | Which is also misleading |
| 22:51 | <csarven> | <div monkey<see="> monkey do</div> <--- How would GRE, WebKit or Presto handle this? |
| 22:55 | gsnedders | thinks a see element as a child of a div element |
| 22:55 | <gsnedders> | But I'm guessing :P |
| 22:56 | <Dashiva> | That's not a well-formedness error, though. That's a batshit-insane-author error. |
| 22:59 | <takkaria> | IIRC it would be a div with a silly attribute and no children |
| 23:06 | <annevk> | takkaria is correct |
| 23:10 | <zcorpan> | hmm. why is the empty string not allowed as id? |
| 23:10 | <zcorpan> | seems convenient to say elm.id = '' instead of elm.removeAttribute('id') |
| 23:11 | <Dashiva> | Jirka brings up a scary point about XSLT-generated |
| 23:11 | <Dashiva> | Wasn't it XSLT-compatible at some point? |
| 23:12 | <zcorpan> | hsivonen suggested XSLT-compat |
| 23:12 | <hober> | I'd prefer something intentionally meaningless and hideously ugly |
| 23:13 | <hober> | like <!DOCTYPE html PUBLIC "f9601f618c1c7a5821c5a4368bd20a17"> |
| 23:13 | <Dashiva> | "I am a muppet who uses XSLT to generate HTML" would never make it to REC |
| 23:13 | <hober> | heh |
| 23:13 | <hober> | (I md5'd "XSLT talisman") |
| 23:14 | <zcorpan> | i like the empty string |
| 23:15 | jgraham | votes for <!DOCTYPE html PUBLIC "papaya-whip" |
| 23:15 | <jgraham> | > |
| 23:15 | <Dashiva> | jgraham: If it's going to be like that, it must at least involve bee-keeping |
| 23:17 | <gsnedders> | jgraham: No, Hixie-rules? |
| 23:19 | <hallvors> | zcorpan: so, would the element be found by document.getElementById('') ? The empty strying isn't equivalent to no attribute. |
| 23:19 | <zcorpan> | hallvors: that's covered in the spec already |
| 23:20 | <hallvors> | (Opera has had real-life incompatibilities with that as far as I remember) |
| 23:20 | <annevk> | it's PapayaWhip |
| 23:21 | <takkaria> | whee, google chrome crashing |
| 23:22 | <takkaria> | though it's quite cool how it didn't take down the browser |
| 23:24 | <Dashiva> | I wonder if they left a few crashers unfixed just so people would see the effect |
| 23:54 | <Hixie> | if you want to see chrome crashing, just go to about:crash |
| 23:54 | <Hixie> | :-) |
