| 00:00 | <annevk> | no html5lib flamewar yet |
| 00:04 | <jgraham> | annevk? |
| 00:06 | <Hixie_> | oh blimey, contentEditable feedback |
| 00:06 | <Hixie_> | 116 e-mails of it |
| 00:06 | Hixie_ | trembles |
| 00:14 | <annevk> | jgraham, over the changes I made |
| 00:17 | <jgraham> | Hixie: OK, mail sent. It may tun out to be inaccurate, I need more sleep. |
| 00:17 | <jgraham> | annevk: I didn' notice anything objectionable. I haven't checked if you broke anything, mind |
| 00:17 | <annevk> | jgraham, btw, if you notice stuff I missed please let me know |
| 00:18 | <annevk> | I did run the tests and made sure they passed |
| 00:20 | <takkaria> | hm, with the introduction of the .at TLD, you can do silly things like "takkaria.⊙ta" |
| 00:22 | <jgraham> | annevk: Some of the non-parser tests seem to be failing (lxml and the sanitizer) |
| 00:22 | <jgraham> | s/lxml/liberal xml/ |
| 00:23 | jgraham | -> bed |
| 00:23 | <annevk> | yeah, didn't check those |
| 00:23 | <jgraham> | annevk: You should send mail to the list saying they fail, at least |
| 00:23 | <annevk> | i did e-mail the list |
| 00:23 | <annevk> | saying I only checked the Python parser |
| 00:24 | <annevk> | if there's nothing there tomorrow i'll mention the rest too |
| 00:24 | <annevk> | (as it's already implicitly there...) |
| 00:25 | <jgraham> | annevk: I just sent mail |
| 00:27 | <annevk> | k |
| 00:28 | annevk | -> bed |
| 02:35 | <jwalden> | zounds |
| 02:35 | <jwalden> | "we recommend that you switch to the Internet Explorer 8 Cross Document Messaging feature that is based on Section 6.4 of the HTML 5.0 specification." |
| 02:35 | <jwalden> | WIN |
| 02:36 | <Hixie> | hm? |
| 02:37 | <jwalden> | http://support.microsoft.com/kb/949787 |
| 02:38 | <Hixie> | jwalden: ah yes. |
| 02:39 | <Hixie> | has anyone found any docs on how they implemented that? (their implementation of DOM storage is quite... embrace-and-extendy.) |
| 02:39 | <jwalden> | there's a note about draft status; I'm betting they have the old interface implemented |
| 02:53 | <roc> | someone should raise a stink about how they've extended storage without giving any feedback (if that's true) |
| 09:07 | <annevk> | hi liorean |
| 09:55 | <annevk> | Hixie, any thoughts on whether it's worth to obsolete DOM Level 2 Views or keep it alive? |
| 10:00 | <annevk> | Philip`, I tested one such gb2312 site you pointed out, <http://www.tkdts.com/>;, and it seems that browsers treat it as UTF-8 |
| 10:01 | <Philip`> | That has "Content-Type: text/html; charset=utf-8" so that would make sense |
| 10:01 | <annevk> | as HTTP header? |
| 10:02 | <Philip`> | Yes |
| 10:02 | <annevk> | http://www.jgbr.com.cn/ has the same |
| 10:03 | <annevk> | (also from the 8 remaining pages) |
| 10:03 | <annevk> | http://www.wuxi-accp.com/ too... |
| 10:04 | <annevk> | http://www.liechebuluo.com/ doesn't |
| 10:05 | <annevk> | few others don't have the same |
| 10:05 | <annevk> | no real pattern |
| 10:10 | <hsivonen> | http://html5.org/tools/web-apps-tracker?from=1254&to=1255 |
| 10:10 | <Philip`> | http://www.liechebuluo.com/ in the left column says "» 奥地利�戴姆勒" in Firefox, "» 奥地利?#25140;姆勒" in IE6, so it looks like the page is simply broken |
| 10:10 | <hsivonen> | does that change have an actual conformance checker-relevant change in it? |
| 10:10 | <hsivonen> | what is it? |
| 10:10 | <hsivonen> | MikeSmith: btw, out=gnu now puts the URI in quotes |
| 10:11 | <MikeSmith> | hsivonen - sweet |
| 10:11 | MikeSmith | wonders if Karl Berry may have updated the GNU coding standards spec yet |
| 10:11 | <hsivonen> | MikeSmith: also, I added an option to use ASCII quotes as requested by Hixie. |
| 10:11 | <hsivonen> | nope |
| 10:11 | <Hixie> | annevk: should probably be merged into whatever does altss and so forth |
| 10:11 | <Hixie> | hsivonen: thanks for that, btw |
| 10:11 | <hsivonen> | asciiquotes=yes |
| 10:12 | <annevk> | Hixie, my idea is to drop the concept of multiple views altogether |
| 10:12 | <Hixie> | greatly appreciated |
| 10:12 | <annevk> | hsivonen, looking for should/must I'd say no |
| 10:12 | <Hixie> | annevk: might want to talk to your engineers first, since opera is the only browser to actually have multiple views :-) |
| 10:12 | <annevk> | Philip`, k |
| 10:12 | <hsivonen> | Hixie: you're welcome |
| 10:12 | <hsivonen> | annevk: thanks |
| 10:12 | <MikeSmith> | hsivonen - I wonder if current Emacs and other apps that have built-in parsing for GNU-format error messages can deal with the quoted filename part as expected |
| 10:13 | <hsivonen> | MikeSmith: I have no idea |
| 10:13 | <annevk> | Hixie, I wonder how much of that is still around... |
| 10:13 | <annevk> | Hixie, but yeah, sure |
| 10:13 | <MikeSmith> | hsivonen - what kind of quotes does it output by default? if not ASCII.. |
| 10:14 | <Hixie> | annevk: does it cause any harm to have it? i think it makes sense to at least specify the theory there |
| 10:14 | <hsivonen> | MikeSmith: aside: I use URI-level quoting instead of C string quoting for " in URI, so the output is more regexp-friendly than the upcoming GNU spec |
| 10:14 | <annevk> | Hixie, so my main concern was actually who'd define the document attribute, but I guess CSSOM View could do that... |
| 10:14 | <hsivonen> | MikeSmith: be default the quotes in messages are U+201C and U+201D |
| 10:14 | <MikeSmith> | ah |
| 10:14 | <annevk> | Hixie, well, all code is written with a single view in mind, hmm |
| 10:15 | <Hixie> | annevk: *shrug* |
| 10:16 | <Hixie> | annevk: i don't really care either way, but if we drop them we should fix UIEvent too |
| 10:16 | <MikeSmith> | hsivonen - when using out=gnu, maybe "yes" should be the default for asciiquotes |
| 10:17 | <MikeSmith> | so you have to set it to "no" if you actually want U+201C and U+201D |
| 10:17 | <MikeSmith> | in out=gnu mode |
| 10:18 | <hsivonen> | MikeSmith: yeah, that would make sense if it only affected quotes. However, it also affects apostrophes in text which in egde cases makes the message technically wrong |
| 10:18 | <MikeSmith> | ah |
| 10:18 | <hsivonen> | MikeSmith: but yeah, I should probably make it the default |
| 10:19 | <hsivonen> | I guess I should make it the default and then make sure that I don't use contractions in the messages, so that I can get rid of the apostrophe thing |
| 10:20 | <MikeSmith> | yeah that might work |
| 10:21 | <annevk> | Hixie, to have .view point to Window all the time? |
| 10:21 | <MikeSmith> | though I wonder if there are some uses of apostrophes in messages that you can't avoid |
| 10:21 | <Hixie> | annevk: to not have .view (since if we only have one view, it'll always point to window) |
| 10:22 | <annevk> | true, but i thought that might break stuff |
| 10:22 | <MikeSmith> | hsivonen - aside: Henry Thompson did a presentation here today about XML pipeline stuff and XML Schema; one thing I learned is that apparently they are adding a mechanism similar to Schematron assertions |
| 10:23 | <MikeSmith> | adding it to the next version of XML Schema |
| 10:23 | <Hixie> | annevk: *shrug*. I don't see much point in removing the multiple views stuff frankly. it doesn't simplify anything really. |
| 10:23 | <hsivonen> | MikeSmith: I've heard something vague about that but I have never seen concrete examples of what XSD is adding |
| 10:23 | <annevk> | Hixie, ok |
| 10:24 | <hsivonen> | MikeSmith: is "here" SXSW? |
| 10:24 | <MikeSmith> | hsivonen - nope, here at home in Tokyo. we had a W3C Japan Member meeting today |
| 10:24 | <MikeSmith> | I presented after Henry |
| 10:25 | <hsivonen> | MikeSmith: ah |
| 10:25 | <MikeSmith> | and used part of my time, to follow up on some of what he discussed, to show validator.nu and to talk about HTML5lib and also the validator.nu parser |
| 10:26 | <MikeSmith> | Henry had mentioned the use case of people wanting to take "bad HTML" and get into a form that could be processed with an XML toolchain |
| 10:27 | <annevk> | Hixie, so should http://dev.w3.org/csswg/cssom-view/#windowview be implemented by the default view AbstractView object? |
| 10:29 | <Hixie> | i think "media" on that interface should be on AbstractView, and the rest should be on ScreenView, which would inherit from AbstractView |
| 10:29 | <Hixie> | it would only be implemented by the defaultView if the defaultView is a screen-media view |
| 10:29 | <Hixie> | or projection media, i guess |
| 10:30 | <Hixie> | or possibly visual media |
| 10:30 | <Hixie> | i dunno what the details would be exactly |
| 10:30 | <Hixie> | btw you have a typo in that idl, sroll -> scroll, twice |
| 10:30 | <annevk> | thanks |
| 10:30 | <annevk> | so I guess i'll port AbstractView to that document |
| 10:31 | <Hixie> | makes sense |
| 10:31 | <Hixie> | is this the doc that does the altss stuff btw? |
| 10:31 | <annevk> | one less DOM Level xxx spec hurray |
| 10:31 | <annevk> | Hixie, no, that's in the far bigger CSSOM spec |
| 10:31 | <Hixie> | ah ok |
| 10:31 | <Hixie> | any eta on that? i have a dependency on it from html5 |
| 10:31 | <annevk> | any eta on HTML5? |
| 10:31 | <Hixie> | not eta to rec or anything |
| 10:32 | <Hixie> | yeah |
| 10:32 | <annevk> | hehe |
| 10:32 | <Hixie> | http://www.whatwg.org/specs/web-apps/current-work/TIMETABLE |
| 10:32 | <annevk> | the main problem is defining getComputedStyle properly |
| 10:32 | <Hixie> | ah yeah that will be fun i'm sure |
| 10:32 | <annevk> | and I have some issues defining the altcss stuff in a way that it is more about a model rather than attribute values |
| 10:33 | <annevk> | but mostly getComputedStyle, serializing and parsing CSS blocks, etc. |
| 10:33 | <Hixie> | k |
| 10:33 | <Hixie> | what kind of issues? |
| 10:34 | <annevk> | one issue is that browsers seem to implement various different CSSStyleDeclaration objects |
| 10:35 | <annevk> | one issue is that getComputedStyle works with CSS 2.0 and not with CSS 2.1 |
| 10:35 | <annevk> | so you can't easily re-use the terminology from CSS 2.1, such as "computed style" |
| 10:36 | <Hixie> | oh i meant about the model thing |
| 10:37 | <annevk> | i think that's mostly because it's complicated |
| 10:37 | <annevk> | i managed to make some progress on it though but haven't looked at it for a while |
| 10:38 | <annevk> | http://dev.w3.org/csswg/cssom/#style |
| 10:39 | <Hixie> | k |
| 10:39 | <Hixie> | well no rush |
| 10:39 | gsnedders | tries to quickly write an email to annevk about XML5 before going to school |
| 10:47 | <gsnedders> | annevk: excuse the likely bad English |
| 10:57 | <Hixie> | hm |
| 10:57 | <Hixie> | how do we prevent people from screwing around with the user selection |
| 10:58 | annevk | thought that was a lost battle |
| 10:58 | <Hixie> | let me rephrase |
| 10:59 | <Hixie> | how do we allow hypothetical user agents that wish to continue fighting that battle win that battle without breaking pages or being non-conforming |
| 11:00 | <annevk> | let the user use a modifier key? |
| 11:03 | <Hixie> | not very discoverable or intuitive |
| 11:03 | <Hixie> | maybe we could say that user agents may ignore API calls to change the selection if they are made immediately after a user has modified the selection |
| 11:03 | <annevk> | the UA could hint about it I suppose |
| 11:04 | <annevk> | that breaks at least one interesting application |
| 11:04 | <Hixie> | which? |
| 11:04 | <annevk> | guys at q42 made a text editor where they implemented selection that was indenting aware |
| 11:05 | <Hixie> | ncie |
| 11:05 | <annevk> | indeed |
| 11:06 | <Hixie> | anyway bed time |
| 11:06 | <Hixie> | nn |
| 11:06 | <annevk> | g'n |
| 11:53 | <xkcd> | Hixie: Since you're a redditor you may want to drop in to this thread: http://reddit.com/info/6b1hq/comments/ |
| 11:53 | <xkcd> | Oh, he's asleep. |
| 12:02 | <Philip`> | It seems slightly inaccurate to say it's "for security reasons", since the reason is mostly that IE chose to implement <object> very differently to how other browsers do it |
| 12:02 | Philip` | doesn't know if anything (like HTML5) specifies that IE is wrong |
| 12:03 | <annevk> | HTML5 doesn't say that cross-domain should fail |
| 12:04 | <zcorpan> | ie8 fails acid2 for me when i allow activex, but passes when i disallow activex (on acid2.acidtests.org) |
| 12:04 | <Philip`> | I guess disabling ActiveX makes it show the <object> fallback content |
| 12:04 | <Philip`> | (which is what it ought to do) |
| 12:04 | <zcorpan> | yeah |
| 12:05 | <annevk> | heh, that makes their explanation of making content available pretty bogus |
| 12:05 | <zcorpan> | i get the same result on webstandards.org/acid2 |
| 12:05 | <zcorpan> | but perhaps i don't have default settings |
| 12:05 | <annevk> | zcorpan, you need www.webstandards.org |
| 12:05 | <zcorpan> | annevk: ah |
| 12:30 | <hsivonen> | the S60 Browser crashes on Hixie's timetable... |
| 12:31 | <zcorpan> | hsivonen: pointer to the timetable? |
| 12:32 | <hsivonen> | zcorpan: http://www.whatwg.org/specs/web-apps/current-work/TIMETABLE |
| 12:34 | <zcorpan> | hsivonen: thanks |
| 12:36 | <hsivonen> | gotta love it when svn manages to miswrite its own log format so that it errs on reading it until the log file is edited manually |
| 13:16 | <Philip`> | hsivonen: What part of SVN reads logs files? |
| 13:17 | <hsivonen> | Philip`: I don't know, but I can't run commit or cleanup until I add the bogus attribute the log file reader wants to see |
| 13:18 | <Philip`> | By "log", do you mean the output of "svn log"? |
| 13:18 | <hsivonen> | Philip`: I mean .svn/log |
| 13:18 | Philip` | can't find a file name .svg/log anywhere on his computer |
| 13:18 | <Philip`> | Uh |
| 13:19 | <Philip`> | .svn/log |
| 13:21 | <hsivonen> | Philip`: the top level of a checkout has it |
| 13:22 | <hsivonen> | the GNU Ælfred2 code just keeps surprising me |
| 13:22 | hsivonen | wants an XML5 SAX parser with a proper iterative tokenizer |
| 13:23 | <Philip`> | Hmm, maybe it's a log for failed transactions, which is why I don't have any such files |
| 13:23 | <Philip`> | (*failed and not yet cleaned up) |
| 13:23 | <annevk> | XML5 is SAX compatible |
| 13:31 | <annevk> | posted about the embrace and extend stuff: http://annevankesteren.nl/2008/03/ie8-bad |
| 13:35 | <Philip`> | annevk: s/reuqests/requests/ |
| 13:35 | <annevk> | thanks |
| 13:36 | <zcorpan> | annevk: pointer to the namespace stuff? |
| 13:37 | <annevk> | as in, you'd like to read about it or it should be in my post? |
| 13:37 | <zcorpan> | the former |
| 13:37 | <zcorpan> | (the latter is up to you) |
| 13:38 | <Philip`> | http://code.msdn.microsoft.com/ie8whitepapers/Release/ProjectReleases.aspx?ReleaseId=573 |
| 13:38 | Philip` | wonders what XPS files are |
| 13:38 | <zcorpan> | Philip`: thanks |
| 13:38 | <annevk> | guess i'll leave it out as i already provided various pointers to http://www.microsoft.com/windows/products/winfamily/ie/ie8/readiness/DevelopersNew.htm |
| 13:39 | <Philip`> | (The namespace stuff appears to not work at all in the current beta) |
| 13:40 | <annevk> | (oh) |
| 13:40 | <annevk> | oh well |
| 13:40 | <annevk> | i thought bits did |
| 13:41 | <Philip`> | I've not found anything that works better than IE7, and http://canvex.lazyilluminati.com/misc/ie8/2.html works worse |
| 13:43 | <annevk> | i guess my comments still stand if it's worse |
| 13:45 | <zcorpan> | it's not clear to me what they have introduced with ie8 wrt namespaces that is incompatible, compared to ie7 |
| 13:45 | <Philip`> | Presumably the beta is just unfinished and buggy, and the whitepaper describes what they hope they'll be able to implement at some point in the future |
| 13:45 | <Philip`> | (preferably without breaking the web) |
| 13:45 | <takkaria> | Philip`: XPS is MS's attempt to overthrow PDF |
| 13:46 | <Philip`> | takkaria: Ah, makes sense |
| 13:46 | <Philip`> | What kind of tools can read it? |
| 13:46 | <takkaria> | Office 2007 |
| 13:46 | <takkaria> | I think they were trying to get printer manufacturers to understand it natively too |
| 13:46 | <Philip`> | Also, shouldn't they not make it obvious that it's 30% less space-efficient than PDF? |
| 13:47 | <takkaria> | it's actually XAML-based |
| 13:47 | <Dashiva> | That just means 30% more awesomeness per file |
| 13:48 | <hsivonen> | PostScript operators are more compact than XML. I guess they still are after deflate |
| 13:48 | <Philip`> | Dashiva: If you made a PDF file that was pure awesome, and then converted to XPS, what would happen? |
| 13:49 | <Dashiva> | The file becomes even more awesome because it's now bigger and can contain more awesomeness |
| 13:49 | <Philip`> | Oh, that makes sense |
| 13:50 | <hsivonen> | Philip`: it becomes XML-based an XML is standard, interoperable and awesome |
| 13:50 | <hsivonen> | s/ an / and / |
| 13:50 | <takkaria> | if you had a never-ending PDF file that was pure awesome, and you converted it to XPS, the two would be equally awesome |
| 13:51 | <Dashiva> | Well, even an infinite signal can have finite energy |
| 13:51 | <Philip`> | Hmm, do PDF or XPS support unending documents? |
| 13:52 | <hsivonen> | they don't |
| 13:52 | <Philip`> | I suppose they'd have a header which lists the number of pages, or something like that |
| 13:53 | <hsivonen> | PDF sure doesn't. it requires a seek to the end of the file in order to start reading |
| 13:53 | <hsivonen> | and XPS is in a PKZIP wrapper, which, I suppose, doesn't support infinite entries |
| 13:54 | <Philip`> | When a zip file spans multiple floppy disks, you have to insert the first disk then the last disk then the first disk again and then the second and third etc |
| 13:54 | <Philip`> | but I don't know if that means zip requires seeking to end, or if it's just the PKZIP implementation that requires it |
| 13:55 | <Philip`> | (Zip stores the list of files at the end, but maybe it has enough repeated data per file to allow non-seeking decompression) |
| 14:23 | <zcorpan> | hmm, having dom attributes for aria might work for legacy browsers anyway, if ATs can read js properties on elements |
| 14:44 | <zcorpan> | Philip`: any idea why aria attributes don't show up in your live dom viewer for ie8? |
| 14:45 | <zcorpan> | i can access them using for-in, and they are .specified |
| 14:45 | <Philip`> | zcorpan: Probably because IE8 is crazy |
| 14:45 | <zcorpan> | no doubt |
| 14:46 | <Philip`> | I use attributes.getNamedItem(name) on them |
| 14:46 | <Philip`> | which is probably what fails |
| 14:46 | <zcorpan> | that works for me |
| 14:46 | <Philip`> | because 'name' is probably 'ariaDisabled' and not 'aria-disabled' |
| 14:46 | <Philip`> | unless I'm confused and/or wrong |
| 14:47 | <zcorpan> | .name is aria-haspopup |
| 14:47 | <Philip`> | Does for-in return "aria-haspopup"? |
| 14:48 | <Philip`> | By the way, where is my DOM viewer? |
| 14:48 | <Philip`> | (I think I've lost it temporarily) |
| 14:48 | <zcorpan> | http://philip.html5.org/misc/live-dom-viewer-ie8.html |
| 14:48 | <Philip`> | Aha, thanks |
| 14:49 | <Philip`> | but I thought I'd deleted that version since http://philip.html5.org/misc/live-dom-viewer-ie8/ has a working permalink button |
| 14:49 | Philip` | deletes it |
| 14:49 | <Philip`> | Please use the non-.html version :-) |
| 14:51 | <Philip`> | Anyway, I'm not quite sure what'd cause that issue, and I don't have access to IE8 to test it right now |
| 14:52 | <zcorpan> | the non-.html version crashes ie8 |
| 14:52 | <Philip`> | Um |
| 14:52 | <zcorpan> | twice, but works now... ? |
| 14:53 | <Philip`> | It's always worked for me |
| 15:02 | <zcorpan_> | hmm, getNamedItem indeed doesn't seem to work |
| 15:03 | <zcorpan_> | it returns null |
| 15:04 | <Philip`> | Would getAttribute work better? |
| 15:14 | <zcorpan_> | only if you reverse the camelcase to hyphenated |
| 15:15 | <Philip`> | Is .attributes meant to have the DOM (hyphenated) attributes? |
| 15:16 | <Philip`> | (in for-in iteration and getNamedItem, at least) |
| 15:17 | <zcorpan_> | for-in returns the camelcase version, getNamedItem or [] notation requires hyphenated |
| 15:18 | <zcorpan_> | that's why it doesn't work |
| 15:21 | <Philip`> | Has somebody noticed that IE8's standards mode is like other browsers' standards mode and never like almost-standards mode? |
| 15:21 | <Philip`> | (which seems kind of bad for compatibility) |
| 15:22 | <zcorpan_> | i guess we need to fix css2.1 before ie8 ships |
| 15:32 | <takkaria> | I wonder why MS seem to not let any of their browser people talk to anyone outside MS about anything |
| 15:33 | <Philip`> | takkaria: That's not entirely true - I think they've posted questions to the HTML WG about three times |
| 15:33 | takkaria | grins |
| 15:37 | <Philip`> | Maybe they'll be willing to talk more now that they've revealed a number of the features they're implementing |
| 15:37 | <takkaria> | I just want to an answer to "why are you implementing APIs async when they're not meant to be?", really |
| 15:39 | <Philip`> | "Internet Explorer 8 Beta 1 for Developers writes items to the store asynchronously so that your Web page can continue on." |
| 15:39 | <Philip`> | Is that not a good enough answer? |
| 15:42 | <Philip`> | It'd be nice to know how begin/commit work if it's spread across multiple script blocks, and is interleaved with execution of another window in the same domain |
| 15:43 | <takkaria> | it's not a bad reason, but "could do better". :) |
| 15:45 | <Philip`> | Acid3 seems to be well timed with the IE8 beta release, since many people are commenting that even though IE8 sort of passes Acid2 it is still not nearly as standards-compliant as other browsers in Acid3 |
| 15:52 | <zcorpan_> | and scores the same as ie7, doesn't it? |
| 15:53 | <Dashiva> | Philip`: sort of? |
| 15:53 | <Philip`> | IE7 gets 13, IE8 gets 17 |
| 16:26 | <tantek> | mmmm.... Las Manitas..... |
| 16:27 | <tantek> | sorry, wrong channel, that was meant for #sxsw ;) |
| 16:51 | <aroben> | annevk: nice blog post (re: http://annevankesteren.nl/2008/03/ie8-bad) |
| 16:57 | <Dashiva> | The second bullet seems a bit vague on what feature they've implemented |
| 16:57 | <othermaciej> | yeah |
| 16:57 | <othermaciej> | I figured out that it is about DOM storage |
| 16:57 | <othermaciej> | but they should say that |
| 18:21 | <Philip`> | Dashiva: It only works on the version hosted on one specific domain, and you could argue that other versions are no less official, hence IE only sort of passes |
| 19:22 | <Philip`> | In Acid2 in IE8, the nose goes blue only when the mouse is above the bottom of the nose, whereas Firefox and Opera make it blue when the mouse is over the line below that |
| 19:22 | <Philip`> | Is that undefined behaviour, or a bug in IE, or not part of the test? |
| 20:37 | <annevk> | hmm, Ubuntu's seemling random freezes are annoying |
| 20:38 | <annevk> | especially when there's IRC history of unlogged channels :( |
| 20:40 | <Dashiva> | annevk: And they still don't have opera 9.26 ;) |
| 20:41 | <annevk> | I know how to get my own builds :) |
| 20:48 | <eseidel> | Philip`: best to ask Hixie about the nose-blue thing |
| 21:11 | <annevk> | grmbl, what's up with my blog |
| 21:12 | <hsivonen> | has the order in which #document-fragment appears in the tests changed? |
| 21:13 | <hsivonen> | it looks like the format no longer matches what I programmed my test harness to expect |
| 21:14 | <hsivonen> | yeah. |
| 21:15 | <hsivonen> | annevk: ok if I reorder the test case fields so that they are in a consistent order |
| 21:15 | <hsivonen> | ? |
| 21:15 | <annevk> | that'd be cool |
| 21:16 | <hsivonen> | ok |
| 21:16 | <annevk> | i'd still like to group them logically somehow at some point |
| 21:21 | <gsnedders> | annevk: remember what the topic is in here :) |
| 21:27 | <Hixie> | xkcd: added a comment, thanks for the headsup |
| 21:31 | <hsivonen> | ooh. there are now serializer tests, too |
| 21:32 | <annevk> | and some fragment stuff too it seems, but looking through the code fragment parsing has lots of bugs |
| 21:32 | <hsivonen> | annevk: do you mean fragment tree building? |
| 21:33 | <annevk> | yeah |
| 21:33 | <hsivonen> | that stuff was around already in the summer |
| 21:33 | <annevk> | true |
| 21:33 | <annevk> | the serializer stuff has been in for quite a while too i think |
| 21:33 | <hsivonen> | hmm. I wonder if I should implement the feed sniffer, too, and whine if people try to validate docs that would be sniffed as feeds |
| 21:45 | <annevk> | http://www.anomalousanomaly.com/2008/03/06/acid-3/ is funny |
| 21:52 | <gavin_> | heh |
| 21:56 | <svl> | weird that 2% gap between gecko 20080306 and 20080305; I only know of the A ~ B C checkin adding an extra point, but where's the other one coming from? |
| 21:56 | <Philip`> | "3.0b3pre nightly (2008030504)" - is that possible? I thought it was 3.0b5pre |
| 21:57 | <svl> | indeed |
| 21:58 | <Dashiva> | No opera 9.5, pssh |
| 21:59 | <gavin_> | Philip`: the date is when it was built |
| 21:59 | <gavin_> | so it's possible that it was a custom build |
| 21:59 | <gavin_> | based on older code |
| 22:02 | <roc> | If Webkit can gain 50% in a few months then once we start trying we'll hit 116%! |
| 22:03 | <roc> | mmm, we should easter-egg that in a nightly build just for laughs |
| 22:04 | <annevk> | hehe |
| 22:04 | <othermaciej> | in a way we gained more than 50% |
| 22:05 | <othermaciej> | cause Hixie kept adding new tests for us to fail |
| 22:09 | <jgraham> | show-off ;) |
| 22:10 | <jgraham> | Will anyone object if I change most of the mixed-case <!DOCTYPE hTmL> things in the html5lib test suite to have lowercase html |
| 22:11 | <hsivonen> | jgraham: I'm not objecting but I'm curious why |
| 22:12 | <jgraham> | The rationale being that the new lxml treebuilder in html5lib doesn't actually store the doctype name (because lxml just stores the root element name instead), so at the moment we have to selectively skip those tests for lxml |
| 22:12 | <annevk> | yeah |
| 22:12 | <annevk> | i do |
| 22:13 | <annevk> | i think having those tests in there is a good thing |
| 22:13 | <othermaciej> | jgraham: Hixie himelf said so |
| 22:14 | <jgraham> | annevk: I agree that testing that odd doctypes work is A Good Thing |
| 22:14 | <othermaciej> | I'm not especially excited about it since it kind of makes Safari 3.0.4 look worse in comparison to other browsers than it actually is, but so it goes |
| 22:14 | <jgraham> | I don't think we need to pollute lots of tests for other markup wih them |
| 22:14 | <annevk> | ok... hmm |
| 22:16 | <jgraham> | (Other solutions include changing the parsing algorithm to downcase the doctype name in the DOM (dunno if that has compat issues) or hacking the test harness somehow) |
| 22:18 | <annevk> | omg |
| 22:19 | <annevk> | why did i walk in the colorblind trap on public-css-testsuite |
| 22:19 | <annevk> | this stuff is hilarious |
| 22:22 | Philip` | doesn't quite see why it is |
| 22:24 | <hsivonen> | jgraham: I agree that the doctype shouldn't have to round trip in data models designed for XML |
| 22:28 | <Dashiva> | annevk: chaals is gonna cricket-bat you if you keep this up ;) |
| 22:29 | <hsivonen> | I think it isn't hilarious, but using color does have a point since it doesn't depend on for renderers, etc. |
| 22:29 | <gsnedders> | Dashiva: I never knew chaals was that kind |
| 22:30 | <Dashiva> | I take it your treatment was worse? |
| 22:30 | <hsivonen> | s/for/font/ |
| 22:30 | <gsnedders> | Dashiva: IE6. |
| 22:31 | <Dashiva> | That's all? |
| 22:33 | <gsnedders> | And chaals, why I make bad joke. |
| 22:33 | <gsnedders> | *while |
| 22:33 | <gsnedders> | *jokes |
| 22:33 | gsnedders | is too tired |
| 22:33 | <hsivonen> | http://www.w3.org/2005/Incubator/rdb2rdf/ interesting |
| 22:40 | <gsnedders> | anyone have any view on which of Computational Physics, Maths with Physics, or Computer Sciences to do at uni |
| 22:42 | <annevk> | ok, not exactly hilarious |
| 22:42 | <annevk> | still, accusations of discrimination when it comes to such a limited target audience seems a bit of stretch |
| 22:44 | <gsnedders> | I mean, does it really matter when a whole ten people in the world will use the test cases? |
| 22:44 | <annevk> | no it doesn't |
| 22:44 | <Philip`> | gsnedders: Is that Computer Science that doesn't include any physics? |
| 22:45 | <gsnedders> | Philip`: no more physics than CS normally does :) |
| 22:45 | <annevk> | there's someone on Opera's QA team who is partially colorblind and he just does different types of tests |
| 22:45 | <gsnedders> | Philip`: and you probably know better than me how much that is |
| 22:45 | <Philip`> | What's normal? :-) |
| 22:45 | <gsnedders> | Philip`: where did you go as an undergraduate? |
| 22:45 | <Philip`> | 25% of my first year was physics |
| 22:45 | <Philip`> | Cambridge |
| 22:46 | <gsnedders> | do you still have to take a Nat. Sci.? |
| 22:46 | <Philip`> | which is not entirely normal |
| 22:46 | <gsnedders> | among the places where I'd even think of doing CS, it is :) |
| 22:47 | <jgraham> | gsnedders: Well Computer Science probably appears on more job adverts, although I wouldn't advocate using that as a metric |
| 22:48 | <gsnedders> | jgraham: most I've seen also say, "or related field", though :) |
| 22:48 | <jgraham> | Physics is only very arguably a related field |
| 22:48 | <Philip`> | gsnedders: As far as I'm aware, you still have to do one sciencey thing (plus maths) in the first year - there are plans to change that so there's a purer CS option, but not for next year (but maybe for the year after) |
| 22:48 | <hsivonen> | gsnedders: I have just two observations: 1) There are people who picked physics but end up doing software-related stuff anyway. 2) Picking physics for the prestige doesn't make sense; one needs to be motivated in order to keep up with the kind of math classes physics majors take. |
| 22:49 | jgraham | hasn't yet found out how big the practical difference is |
| 22:49 | <gsnedders> | Philip`: I wouldn't be starting until 2009 anyway |
| 22:50 | <Philip`> | gsnedders: Ah, in that case I think it's quite possible that there's be a non-natsci option, but I don't know what new material they'd introduce |
| 22:50 | <Philip`> | s/'s/'d/ |
| 22:50 | <gsnedders> | hsivonen: 1) I'm well aware, which is part of the reason why I am considering physics with maths at all; 2) I'm probably more interested in physics than I am in CS. |
| 22:50 | <jgraham> | gsnedders: With Cambridge you can do 50% C.S., 25% Maths and 25% Physics in the first year and choose Physics or C.S. in the second year |
| 22:51 | gsnedders | will probably nag jgraham and Philip` about Cam. more in May |
| 22:52 | <jgraham> | gsnedders: Do you have a long term goal e.g. academia, ordinary employment, start own business, etc. |
| 22:52 | <jgraham> | ? |
| 22:53 | <gsnedders> | jgraham: NOT academia (that's the one result of having a father in that). :) Some sort of CS-related job, likely either programming or QA initally |
| 22:53 | <Philip`> | The first year is usually not terribly interesting anyway, since it's largely covering things that half the people already know half of, and after that there's no mixing between CS and other departments |
| 22:54 | <hsivonen> | gsnedders: fwiw, the math bit is from experience. Where I studied, one had to test higher to get into physics than CS. I wanted to do CS, but since on paper I was smart enough to do physicist math, I took physicist math inititally out of vanity. It wasn't a smart choice, since I didn't really care enough about hard math and the CS kind of math would have been more useful to me up front |
| 22:55 | gsnedders | would likely go straight into second year if he went to Edi., as I'll have done an extra year at school over the Scottish norm, which has the fun of not being taught some of the more specific stuff and then have to catch up quickly |
| 22:56 | <Hixie> | annevk: when people complained to me, i pointed out that people with visual disabilities are probably not ideally suited to doing visual quality assurace |
| 22:56 | <gsnedders> | hsivonen: Comparing where the two places I'd really lean towards between the two, the entrance requirements are more or less identical, though CS may be harder to get into. I don't really care about vanity. |
| 22:56 | <jgraham> | gsnedders: As someone who has latterly decided that academia is a bad idea I am finding that I would have benefited from more CS in my background, either from classes or from personal learning |
| 22:56 | <Philip`> | (Incidentally, I think there are plans for an optional fourth year in the CS tripos here, which should be introduced soonish - I'm not sure if that's mentioned in any prospectus-like documents yet) |
| 22:57 | <gsnedders> | jgraham: heh. I forget what you do? |
| 22:57 | <jgraham> | gsnedders: I'm coming to the end of an astrophysics PhD |
| 22:58 | <gsnedders> | jgraham: you're postgrad too? oh. My memory really is bad. |
| 22:58 | <Hixie> | personally i've found that having the scientific method drilled into me was very useful, and that i learnt pretty much all the CS i needed for my career on my own time |
| 22:59 | <hsivonen> | gsnedders: well I'm not sure if vanity is the right way of describing it. I was silly enough to think that it would be a good idea to take the "guru" math since I was eligible. |
| 22:59 | <othermaciej> | physics is a great major if you want a high-paying job in finance |
| 23:00 | <gsnedders> | All the CS I've needed for what I've done I've more or less learnt myself or found people around on the web (recently mainly here). There's very little I haven't managed to get my head around myself. |
| 23:00 | <gsnedders> | hsivonen: Mini-story: I was given a choice between being in the top set, or in the third set (of eleven), for Maths. I chose the third set, not wanting to act like a know-it-all. |
| 23:00 | <jgraham> | othermaciej: And you can stomach the hours, the work environment, the lack of ethics... |
| 23:01 | <gsnedders> | othermaciej: finance would kill me. :) |
| 23:03 | <gsnedders> | hsivonen: re: entrance requirements: maths & phys and CS have identical requirements (comparing ed and cam respectively) |
| 23:03 | <Philip`> | I think there are at least a few things I probably wouldn't have learnt myself, like Fourier transforms and discrete maths and Turing machines and lambda calculus and FSMs and logic and semantics and security and quantum computing and stuff |
| 23:04 | <Philip`> | mainly because I wouldn't have any motivation to learn about those things myself |
| 23:04 | <blooberry> | hixie: I remember that email thread you had about color blindness. 8-} |
| 23:04 | <Philip`> | because they're not always entirely useful; but there are always a few occasions when I'm glad I know about such things :-) |
| 23:04 | <jgraham> | FWIW out of those I did Fourier transforms and Quantum computing in a physics degree |
| 23:05 | <Philip`> | I was taught about Fourier transforms three times, and at the end I almost understood them |
| 23:05 | <gsnedders> | Philip`: I nagged Turing machines out of you, IIRC. |
| 23:05 | <gsnedders> | Big-O notation is one of the few things I really don't get my head around |
| 23:05 | <gsnedders> | (and don't try now) |
| 23:06 | <Philip`> | gsnedders: Knowing what a Turing machine is is not the same as knowing how you can prove that you can't write a program to calculate how much memory a Turing machine is going to use |
| 23:06 | <gsnedders> | More than anything, it's between Comp. Phys. and Comp. Sci., FWIW |
| 23:07 | <Philip`> | (Not that that's a particularly useful thing to prove...) |
| 23:07 | <gsnedders> | Philip`: what a horrible thing to prove |
| 23:07 | <gsnedders> | Philip`: But yeah, realistically, how many times will I need to know that? :P |
| 23:08 | <Hixie> | well, actually, the halting problem came up in the web forms 2 vs xforms transitional thing |
| 23:08 | <Hixie> | so... |
| 23:08 | <gsnedders> | but actually prove it? |
| 23:08 | <roc> | A man^H^H^Hcomputer's got to know his limitations |
| 23:08 | <Hixie> | well you don't need to actually prove it more than once, but you do need to know the proofso that you know whether it can be applied to another situation or not |
| 23:09 | <Philip`> | It's useful to know that you can't decide any interesting property of programs |
| 23:09 | <jgraham> | Oddly enough Dave Raggett has a Physics degree |
| 23:09 | <roc> | actually that's not true |
| 23:09 | <Hixie> | understanding how something works is a very important component of using something |
| 23:09 | <Hixie> | roc: what isn't? :-) |
| 23:09 | <roc> | there are entire workshops devoted to proving termination of programs |
| 23:10 | gsnedders | has a very crude idea of how to prove it |
| 23:10 | <Philip`> | Microsoft has a useful termination-prover for device drivers |
| 23:10 | <Philip`> | but it's not deciding termination |
| 23:10 | <roc> | you can decide for some programs |
| 23:10 | <Philip`> | since sometimes it can only tell you that it can't prove termination |
| 23:11 | <roc> | maybe even almost all programs |
| 23:11 | <Hixie> | Philip`: many web browsers have a halting problem solver too. usually implemented as a timeout. :-P |
| 23:11 | <gsnedders> | how technically complex! :P |
| 23:13 | <Philip`> | roc: That's theoretically useless if it doesn't work for any arbitrary program :-) |
| 23:13 | <gsnedders> | roc: when will I terminate? |
| 23:13 | <gsnedders> | roc: will I> |
| 23:13 | <gsnedders> | s/>/?/ |
| 23:13 | <roc> | gsnedders: I can compute an upper bound, which is considered a satisfactory solution to this problem |
| 23:14 | <gsnedders> | roc: thx |
| 23:14 | <roc> | Undecidability results, like NP-completeness results, can be misleading because in practice a lot of those problems are only "really hard" on a very small subset of the instances |
| 23:16 | <roc> | e.g. in the last decade or two, there was in an explosion in the construction of SAT solvers and their application to lots of different domains, because although SAT is NP-complete almost all SAT problems, even huge ones, are relatively easy |
| 23:16 | <Philip`> | I've always wondered why people care about decidability of type systems, since you're always going to run the program after you've compiled it and then you've got no hope of always terminating |
| 23:16 | <roc> | keeps the build times down |
| 23:17 | <jgraham> | What's a SAT? |
| 23:17 | <Philip`> | You can still get exponential (I think) build times in something like ML, so it doesn't guarantee that the compiler won't take longer than the lifetime of the universe |
| 23:17 | jgraham | discovers google |
| 23:17 | <roc> | A SAT problem is basically "find a set of assignments to boolean variables that makes this huge boolean formula true: ..." |
| 23:17 | <gsnedders> | Philip`: ML? |
| 23:18 | <roc> | Philip`: that's true. In fact the more advanced type theorists have given up on decidable type theories :-) |
| 23:18 | <Philip`> | gsnedders: The language |
| 23:18 | <Philip`> | GHC has some extensions to Haskell that apparently make its type system Turing complete, which is kind of crazy |
| 23:19 | <gsnedders> | I'm going all Web 2.0 with my computing coursework: "Processr" |
| 23:19 | <gsnedders> | Philip`: the _type system_ is Turing complete? That's mad. |
| 23:22 | <Dashiva> | gsnedders: Any worse than C++ templates, or preprocessor macros? |
| 23:23 | <Philip`> | The C preprocessor isn't Turing complete since it can't do anything like recursion |
| 23:23 | <Philip`> | C++ templates are mad :-) |
| 23:23 | <gsnedders> | Dashiva: Ah, that's true. You do have that, but seemingly more useful |
| 23:23 | <gsnedders> | (with the preprocessor, at least) |
| 23:23 | <gsnedders> | (I know nothing about C++ templates) |
| 23:25 | gsnedders | ought to learn C++ |
| 23:25 | gsnedders | has said that before |
| 23:26 | <takkaria> | people use C++ as an excuse to perform frankly inexcusable deeds |
| 23:28 | <Dashiva> | takkaria: Much like everything else man comes in contact with |