| 00:19 | <Philip`> | Hixie: "The strokeRect() method must draw stroke the specified rectangle's path" - s/draw stroke/something/ |
| 00:32 | <Hixie> | "The strokeRect() method must something the specified rectangle's path" ? :-) |
| 00:32 | <Philip`> | Argh, my automatic spec-annotating script can't handle "Bézier" |
| 00:33 | <Philip`> | Hixie: Sounds good to me! |
| 00:37 | <Philip`> | Hmm, can I complain that Opera's ImageData.data isn't a JS array? |
| 00:37 | <Philip`> | at least to the extent that ImageData.data.join(',') doesn't work |
| 00:38 | <Philip`> | (where "complain" means "add a test case for", preferably with some kind of justification from specifications, except I can't see anything saying what IDL "int[]" means...) |
| 00:42 | <Hixie> | Philip`: i'm waiting for dom bindings for js spec before really going there |
| 00:43 | <Philip`> | "The current path, shadow attributes, global alpha, clipping path, and global composition operator must not affect the getImageData() and putImageData() methods." - should that mention the CTM too? |
| 00:44 | <Philip`> | Hixie: Okay |
| 00:44 | Philip` | will add a test anyway, because it causes interoperability problems in practice |
| 00:44 | <Hixie> | i thought i'd mentioned CTM with imagedata already |
| 00:44 | <Philip`> | I can't see it mentioned anywhere |
| 00:45 | <Hixie> | seems not indeed |
| 00:54 | <Philip`> | Hixie: "The ImageData object's width is greater than zero." (and subsequent lines) is wrong, since it's talking about an object that's explicitly not an ImageData |
| 01:06 | <Philip`> | Hixie: What happens with NaN in ImageData.data? (It's not below 0, or above 255, and it can't be rounded to the nearest integer) |
| 01:12 | <Hixie> | Philip`: i'm about to go offline for a few days, can you mail the last two lines you wrote to the list/me? |
| 01:12 | <Philip`> | Hixie: Okay, will do |
| 01:12 | <Hixie> | (well, not completely offline, but likely not working on the spec) |
| 01:12 | <Hixie> | thanks dude |
| 01:13 | <Hixie> | your help is, as ever, highly appreciated |
| 01:17 | <Philip`> | Hmm, is NaN a Number? |
| 01:18 | <AwayEagle> | in ES sense, yes |
| 01:18 | <Philip`> | Ah, thanks |
| 01:18 | <othermaciej> | it's not an instance of the Number builtin class, but it is an intance of the Number value type |
| 01:18 | <Philip`> | (That is mildly confusing...) |
| 01:18 | <othermaciej> | (to be exactingly clear) |
| 01:18 | <Philip`> | Oh |
| 01:18 | <othermaciej> | what, that Not a Number is a Number? |
| 01:19 | <Philip`> | Yes |
| 01:19 | <Philip`> | I'm mostly just wondering if "a data attribute whose value is an enumerable list of values that are either JS Numbers or the JS value undefined." is allowed to contain NaN |
| 01:19 | <Philip`> | so it sounds like it probably is |
| 01:19 | <AwayEagle> | I don't remember, is the Number constructor object a Number? :-) |
| 01:19 | <othermaciej> | in ECMA-262 terms, it's a 4.3.20 Number Type |
| 01:20 | <othermaciej> | not a 4.3.21 Number Object |
| 01:20 | <Philip`> | (or at least it's close enough to be worth being explicit about how it's handled) |
| 01:20 | <othermaciej> | sorry, I meant it's a 4.3.19 Number Value (NaN that is) |
| 01:20 | <othermaciej> | NaN is a JS Number, yeah, in that sense |
| 01:20 | <AwayEagle> | you could use 4.3.20 and 4.3.23 too ;-) |
| 01:20 | <Philip`> | Is e.g. 7 the same 4.3.19 Number Value too? |
| 01:20 | <othermaciej> | to be anal it could distinguish Number Type and Number Object |
| 01:21 | <othermaciej> | Philip`: yes |
| 01:21 | <Philip`> | Okay |
| 01:21 | <othermaciej> | the Number Object type only comes into play when you call methods on a number or explicitly invoke the Number constructor to make a wrapper object |
| 01:21 | <othermaciej> | primitive types in ECMAScript (Number, String, Boolean) have corresponding wrapper object types |
| 01:21 | <Philip`> | That makes sense |
| 01:22 | <othermaciej> | which are sometimes created automatically |
| 01:22 | <AwayEagle> | othermaciej: oh, I need to show you the nasty String.prototype.toString testcase I showed ggaren |
| 01:22 | <Philip`> | Similar to Java, I guess? |
| 01:22 | <othermaciej> | Philip`: yes, but worse |
| 01:26 | <AwayEagle> | othermaciej: http://rafb.net/p/jLOXKL92.html (speaking of worse) |
| 01:28 | <Philip`> | Hmm, Opera 9.5 has gone from 93% down to 81% on the ImageData tests |
| 02:04 | <heycam> | Hixie, for an acid3 submission, what's the best way to get a new document? create an iframe and insert it into the document? |
| 02:05 | <heycam> | or should i use the getTestDocument() function? |
| 02:40 | <Hixie> | heycam: just use createDocument on the DOMImplementation object |
| 02:40 | <Hixie> | heycam: getTestDocument() is used too much already |
| 02:40 | <Hixie> | heycam: and creating an iframe is asynchronous and so wouldn't work in the function |
| 02:57 | <takkaria> | gsnedders: I don't know how much data you have on the cookie5 issue, but |
| 02:58 | <takkaria> | gsnedders: http://users.ecs.soton.ac.uk/jmb/test/cookies.php yields the results shown in http://users.ecs.soton.ac.uk/jmb/test/results/cookies.txt for a bunch of major browsers |
| 08:33 | <MacDome> | Hixie: I need you to explain your @font-face test at some point |
| 08:33 | <MacDome> | I dont' see why your X is in the right place |
| 08:34 | <Hixie> | the containing block is the rectangle the size of the viewport, anchored at the origin |
| 08:34 | <MacDome> | Hixie: I guess the 20px is supposed to be measured from the top of the viewport instead of the top of the body? |
| 08:34 | <Hixie> | yes |
| 08:34 | <Hixie> | the body isn't position:relative or anything |
| 08:34 | <MacDome> | interesting |
| 08:34 | <MacDome> | I wonder if we get that wrong for all absolute positioning |
| 08:35 | <MacDome> | I don't think generated content would be treated differently here |
| 08:37 | <MacDome> | Hixie: well, I'll have to ask hyatt |
| 08:37 | <Hixie> | heh |
| 08:38 | <MacDome> | oh, actually |
| 08:38 | <MacDome> | hum... we seem to do this right most of the time |
| 08:38 | <MacDome> | why fail in this case? |
| 08:40 | <MacDome> | Hixie: and he margins on <html> don't affect this? |
| 08:40 | <MacDome> | or on :root ? |
| 08:41 | <Hixie> | why would they? |
| 08:41 | MacDome | doens't know |
| 08:41 | MacDome | hasn't read css 2.1 yet :) |
| 08:42 | MacDome | kinda thinks of <html> as the viewport, but maybe its' not |
| 08:42 | <MacDome> | Hixie: yeah, margins on <html> seem to affect absolute positioning in Safari and FF |
| 08:42 | <MacDome> | your test would suggest that they're not supposed to |
| 08:43 | <Hixie> | see css2.1 section 10.1 |
| 08:44 | <Hixie> | first bullet point |
| 08:47 | <MacDome> | it's actually bullet 4 which is most intersting |
| 11:30 | <hsivonen> | Can someone remind me why Unicode reserves U+FFFE as a non-character? |
| 11:33 | <annevk> | maybe for bitshifting? |
| 11:36 | <hsivonen> | annevk: oh right. it's the byteswapped FEFF |
| 11:36 | <hsivonen> | silly me |
| 11:37 | <hsivonen> | thanks |
| 12:14 | <zcorpan> | annevk: i think the 512 bytes thing for the <meta charset> is since long gone |
| 12:38 | <annevk> | fixed |
| 12:44 | <hsivonen> | it is interesting how the mindset in some circles has shifted from everything-in-XML to everything-in-JSON |
| 12:50 | <othermaciej> | I insist on everything in octet sequences |
| 12:50 | <othermaciej> | it's what the link layer demands! |
| 12:51 | <zcorpan> | "These elements have new meanings in HTML 5 which are incompatible with HTML 4. The new meanings better reflect the way they are used on the Web or gives them a purpose so people can start using them." |
| 12:51 | <zcorpan> | i find this paragraph confusing and i think it gets the wrong point across |
| 12:51 | <zcorpan> | the point it gets across is that html5 is incompatible with html4 |
| 12:52 | <zcorpan> | i'd suggest to change it to something along the lines of "These elements have slightly modified meanings in HTML5 to better reflect how they are used on the Web or make them more useful." |
| 12:54 | <zcorpan> | also, the last bullet point doesn't seem to fit in that section. it discusses the optionalness of an attribute -- not changed meanings |
| 12:56 | <zcorpan> | i'd suggest to either strike it or phrase it in such a way that it discusses the meaning of <img> when it has no alt='' |
| 13:01 | <zcorpan> | since <img> with no alt='' represents an image and <img> with alt='' also represents an image, it seems that the meaning hasn't really changed since html4 |
| 13:33 | <jwalden> | zcorpan: note Hixie's out for a few days and not watching, probably want to send email |
| 13:34 | <zcorpan> | jwalden: this is about the differences document, which annevk is editing |
| 13:34 | <jwalden> | ah |
| 13:34 | <jwalden> | carry on! |
| 13:34 | <zcorpan> | :) |
| 15:32 | <SadEagle> | Hixie: hi... are you perhaps here and have a second free? |
| 16:04 | <jwalden> | SadEagle: he's gone for a couple days, send email |
| 16:06 | <SadEagle> | jwalden: thanks. does he have a preferred labeling scheme or something? |
| 16:06 | <jwalden> | "labeling scheme"? |
| 16:07 | <SadEagle> | as in titles, etc, to help him sort things out. |
| 16:09 | <jwalden> | dunno |
| 16:12 | <Philip`> | SadEagle: For emails? Things get sorted into folders like http://www.whatwg.org/issues/ manually, so the subject line doesn't seem critical |
| 16:13 | <SadEagle> | Philip`: thanks. is there an acid3 issues page, perhaps? |
| 16:19 | <Philip`> | SadEagle: There isn't - I'd guess most Acid3 discussion is by IRC or private email |
| 16:21 | <jwalden> | so far I've found he's been responsive enough here to suffice |
| 17:21 | <annevk> | zcorpan, fixed |
| 18:24 | <jgraham> | Does anyone have a good example of a table where colgroups are needed, assuming that table headers naturally apply to all columns they span? |
| 18:27 | <jgraham> | (specifically I'm looking at the headers thing again, I understand that, in principle, colgroup can be used for other things) |
| 18:31 | <jgraham> | Presumably such a table would have a header that didn't span an entire colgroup but did apply to all the cells in the colgroup. |
| 20:04 | Philip` | installs KDE 4.0.0 |
| 20:04 | <takkaria> | .0.0 releases make me nervous |
| 20:04 | <Philip`> | Now I can sort of run the canvas tests in Konqueror, except half of them say "Service Temporarily Unavailable" |
| 20:04 | <Philip`> | takkaria: That's why I'm not replacing KDE3.5 with it, and am running it under a separate user account :-) |
| 20:05 | <takkaria> | ah, good move. :) |
| 20:09 | <SadEagle> | Philip`: cool. But if you're building from source, you're better off using 4.0 branch. I fixed a bunch of whoppers on there |
| 20:09 | <SadEagle> | Philip`: and fredrikh made some canvas progress since then. I can give you my test results, but they're kind of cheating |
| 20:12 | <Philip`> | SadEagle: I'm just using Gentoo's packages for now - would it be straightforward to just compile Konqueror from source, without having to manually build the rest of KDE4? |
| 20:12 | <SadEagle> | depends on your definition of straightforward and the rest. You'd only need kdelibs. But I don't know if gentoo does any tricks for concurrent installs and whatnot |
| 20:14 | <SadEagle> | Philip`: do you have some sort of an AVM router? |
| 20:16 | <Philip`> | SadEagle: "straightforward" means copying-and-pasting a few commands to get it compiled and installed somewhere safe (like in /home since I don't want to mess up anything on my system) and running, ideally |
| 20:16 | <Philip`> | AVM router? |
| 20:18 | <SadEagle> | german(?) router vendor of some sort.. a lot of people reporting the timeout problem seem to have one of those. |
| 20:19 | <Philip`> | Ah |
| 20:19 | <Philip`> | I don't, and it was a server-side error message anyway |
| 20:19 | <SadEagle> | As for straightforward.. hmm, it's not too few commands, but there are howtos that are very safe, let me dig it up. Actually, if you're building with tests on, you only need kdelibs, though testkhtml isn't much of a browser :-) |
| 20:20 | <Philip`> | On a page with loads of iframes, is there some limit on how many concurrent connections Kwhatever will create? |
| 20:20 | <Philip`> | (I saw dozens of kio-something processes, which seemed odd) |
| 20:20 | <SadEagle> | not sure. I need to double check. there may be too many :( |
| 20:21 | <SadEagle> | note that there is a bug with onload events with many iframes using the same image -- that's the 'cheating' part I was referring to |
| 20:21 | <Philip`> | Ah |
| 20:22 | <Philip`> | Incidentally, please don't trust my tests to be correct if they're not saying something that's justified by the spec, since I haven't carefully reviewed them at all :-) |
| 20:23 | <SadEagle> | it should limit it to 32 requests, in theory, anyway |
| 20:23 | <Philip`> | Hmm, 32 seems a bit much when most browsers limit it to about 2... |
| 20:24 | <SadEagle> | that's global for the entire process, though. |
| 20:24 | <SadEagle> | may be it needs some sort of fancier policy |
| 20:25 | Philip` | should run the tests on his local machine instead of potentially killing html5.org |
| 20:25 | SadEagle | tests whether the limit really works. |
| 20:26 | <SadEagle> | oh. |
| 20:26 | <SadEagle> | there might be a problem with iframes without a content-type specified :( |
| 20:27 | <SadEagle> | wrt to # of requests. |
| 20:27 | <Philip`> | How do you specify a content-type on an iframe? |
| 20:32 | <SadEagle> | good point. only doable on object, right? :( |
| 20:33 | <Philip`> | Apparently so |
| 20:33 | <Philip`> | I used <iframe> rather than <object> because FF3-trunk didn't seem to do onload on <object> at the time when I was testing it |
| 20:33 | <SadEagle> | this may be hard to fix, will have to talk to some kparts gurus, better run it locally then. |
| 20:37 | <Philip`> | It's surprising how many bugs can be found in a test harness, without even considering the tests themselves :-) |
| 20:38 | <Philip`> | s/in/by/ |
| 20:41 | <SadEagle> | the easiest way to build from source safely is to use this script, and not as .bashrc -- but an explicit one http://techbase.kde.org/index.php?title=Getting_Started/Increased_Productivity_in_KDE4_with_Scripts/.bashrc |
| 20:41 | <SadEagle> | then one can just source it in a terminal window for KDE4 stuff, and run things from there.. |
| 20:43 | <SadEagle> | most likely working steps would be: 1) cd 2) mkdir kde 3) cd kde 4) download that script as ~/kde4/setup-env 5) mkdir src 6) mkdir build |
| 20:45 | <SadEagle> | 7) cd src 8) svn checkout svn://anonsvn.kde.org/home/kde/branches/KDE/4.0/kdelibs 9) edit that setup-env script to remove the alias make=makeobj line 10) cd ../build && mkdir kdelibs && cd kdelibs 11) source ~/kde/setup-env 12) cmakekde |
| 20:45 | <SadEagle> | chances of me not making a mistake there: 0. |
| 20:46 | <SadEagle> | actially, step 10 is not needed :-) |
| 20:47 | <Philip`> | Thanks! |
| 20:47 | Philip` | tries that |
| 20:47 | <SadEagle> | thanks for doing this |
| 20:53 | <Philip`> | "CMake Error: The source directory "/home/kde4/kde/src" does not appear to contain CMakeLists.txt." |
| 20:53 | <Philip`> | when doing step 12, having skipped step 10 |
| 20:54 | <SadEagle> | cd ~/kde4/src/kdelibs, sorry. |
| 20:55 | <Philip`> | That seems happier - thanks again :-) |
| 20:56 | <SadEagle> | and thanks again for testing. Except now I'll have to fix the multi-frame image loading/onload bug ;-) |
| 21:23 | <Philip`> | Hmm, now I'm running something that might be the newly compiled version though I'm not certain |
| 21:23 | <SadEagle> | well, it didn't build konqueror, so probably not :-) |
| 21:23 | <SadEagle> | unless you also build kdebase when you weren't looking |
| 21:24 | <SadEagle> | what it did build is testkhtml.. go to ~/kde/build/kdelibs/bin, ./testkhtml <url goes here> |
| 21:24 | <Philip`> | I told it to run the old (4.0.0) Konqueror with the new ~/kde/lib directory |
| 21:25 | <Philip`> | and it prints lots of debug-like messages to the console, which it doesn't do when I run without ~/kde/lib |
| 21:25 | <Philip`> | so it sounds like it's doing something different |
| 21:26 | <SadEagle> | hmm, I am honestly not sure what will happen. easy way to check, though: cat /proc/<pid>/maps | grep khtml, and see where it comes from |
| 21:26 | <Philip`> | (Incidentally, Canvex seems to reveal a bug - the upscaling image filter wraps around the edges) |
| 21:26 | <SadEagle> | hmm, odd. |
| 21:27 | <Philip`> | b456b000-b4a3a000 r-xp 00000000 08:04 8930658 /home/kde4/kde/lib/libkhtml.so.5.0.0 |
| 21:27 | <SadEagle> | sounds like it worked. |
| 21:27 | <Philip`> | (Also, Canvex reveals an apparent lack of XHTML support :-p ) |
| 21:27 | <Philip`> | Also, View Source crashes |
| 21:28 | <SadEagle> | Philip`: for reasons I am too embarassed to admit. it's using the html parser. Actually, I completely forgot why we didn't switch it over. |
| 21:28 | <Philip`> | (but maybe it's not helped by me having a mixture of different KDE versions) |
| 21:30 | <Philip`> | I think parsing XML with an HTML parser is sadly non-conforming :-( |
| 21:31 | <SadEagle> | :( |
| 21:33 | <SadEagle> | one of those days I'll understand XHTML and flip the switch. |
| 21:35 | <Philip`> | Failed assertion ctx.fillStyle matches /^#(?!(FF0000|ff0000|f00)$)/ (got #ff0000[string]) |
| 21:35 | <Philip`> | How come that doesn't match? |
| 21:36 | <Philip`> | Oops |
| 21:36 | Philip` | missed the ! |
| 21:38 | <Philip`> | Also, it'd be nice to have getImageData so the test results are more automatic :-) |
| 21:38 | <SadEagle> | yeah, if I can get Hixie to make it spec'd in a way I can do it w/o killing the GC |
| 21:51 | <Philip`> | Once this is all complete, you'll have to start on 3D canvas ;-) |
| 21:51 | <SadEagle> | that's for fredrikh, not me :-) |
| 22:42 | Philip` | discovers that his bank now allows Opera, instead of redirecting to a page which supposedly lists supported browsers but actually they obviously didn't test that page since it didn't work |
| 22:53 | <jruderman> | hah |
| 23:52 | <Philip`> | (Ooh, Safari on Wine - that's quite useful) |
| 23:54 | <kig> | http://dark.fhtr.org/safari_on_wine.jpg and then it crashes D: |
| 23:56 | <Philip`> | It crashed for me before I installed all the corefonts, but now it seems happy enough |
