| 02:03 | <MikeSmith> | kennyluck: I'm on my way in to SFC |
| 02:04 | <MikeSmith> | I guess I'll get there by 11:15 |
| 02:04 | <MikeSmith> | so we could meet up for lunch |
| 02:26 | <kennyluck> | MikeSmith: I'll be there by 11:30 |
| 02:26 | <MikeSmith> | kennyluck: cool |
| 02:27 | <MikeSmith> | kennyluck: it's cooled down a lot today |
| 02:27 | <MikeSmith> | because of the rain I guess |
| 03:51 | <MikeSmith> | how do I get OSX to reload my /etc/hosts file? |
| 03:52 | <wirepair> | hu? |
| 03:52 | <wirepair> | is your nsswitch.conf set to use hosts file? (it should. dunno why it wouldn't) |
| 03:53 | <wirepair> | should say like hosts: files dns |
| 03:53 | <MikeSmith> | ok |
| 03:53 | <MikeSmith> | lemme describe what I actually want to do |
| 03:53 | <MikeSmith> | I want to make an alias for localhost |
| 03:54 | <MikeSmith> | what's the right way to do that? |
| 03:54 | MikeSmith | checks nsswitch.conf |
| 03:54 | <wirepair> | edit hosts to have a line like: localhost buh.not.localhost.com |
| 03:54 | <MikeSmith> | locate nsswitch.conf .. しーん |
| 03:55 | <MikeSmith> | ok, done that |
| 03:55 | <MikeSmith> | but how to I get the OS to actually re-read that |
| 03:55 | <wirepair> | it should be instant |
| 03:55 | <MikeSmith> | it's not :( |
| 03:56 | <wirepair> | weird... try like 127.0.0.1 something |
| 03:56 | <wirepair> | then ping someting and see if you get a response? |
| 03:56 | <MikeSmith> | I have no nsswitch.conf btw |
| 03:56 | <wirepair> | weird thought osx would have it ;) |
| 03:57 | <wirepair> | http://forums.macrumors.com/archive/index.php/t-614816.html |
| 03:57 | <wirepair> | try that? |
| 04:20 | <roc> | annevk: Cameron starts in a few weeks |
| 05:33 | <MikeSmith> | wirepair: that link appears to be a couple of confused dudes further confusing each other |
| 05:46 | <nimbupani> | it is ploughman no MikeSmith :P |
| 06:50 | <nattokirai> | MikeSmith: ping |
| 06:53 | <hsivonen> | https://bugzilla.mozilla.org/show_bug.cgi?id=593963#c11 |
| 07:23 | <annevk> | how do you get file size listing again in shell? |
| 07:25 | <hsivonen> | ls -al |
| 07:26 | <hsivonen> | http://www.w3.org/TR/webarch/summary makes it easy to see how disproportionalely "Web" arch is about XML |
| 07:36 | <cyberix> | Is there a concensus about using IPv6 addresses in urls? |
| 07:38 | <MikeSmith> | nattokirai: hey man |
| 07:38 | <MikeSmith> | back now |
| 07:39 | <cyberix> | Would http://[2001:200:dff:fff1:216:3eff:feb1:44d7]/ work on all modern browsers? |
| 07:39 | <MikeSmith> | nattokirai: long time no talky |
| 07:40 | <kennyluck> | cyberix: it works in Safari and Firefox at least |
| 07:40 | <cyberix> | Seems to work on Chrome |
| 07:41 | <variable> | cyberix, likely |
| 07:41 | <cyberix> | seems to work in Opera |
| 07:41 | <variable> | for some definition of modern |
| 07:41 | <cyberix> | Does anyone have an IE around? |
| 07:42 | <variable> | cyberix, fails to work on links |
| 07:42 | <variable> | but I'm not sure if I compiled it with IPV6 support ;-) |
| 07:45 | <cyberix> | variable: :-) |
| 08:06 | <hsivonen> | someone should make a TTY graphics layer for Gecko or WebKit to replace Lynx and Links |
| 08:21 | <annevk> | complete/ is almost resolved |
| 08:21 | <annevk> | meanwhile http://html5.org/complete/ still functions |
| 08:23 | <hsivonen> | I wonder if we already have a test case that checks if http://html5.org/tools/web-apps-tracker?from=5296&to=5297 has been implemented |
| 08:36 | <zcorpan_> | hsivonen: is argument.callee disallowed in strict mode? |
| 08:36 | <zcorpan_> | s//s/ |
| 08:38 | <zcorpan_> | seems so (at least in gecko) |
| 08:41 | <hsivonen> | zcorpan_: I don't know |
| 08:48 | <hsivonen> | apparently we don't have a test for spec rev 5297 |
| 09:00 | <hsivonen> | hmm. I wonder how to write a test case for rev 5297 |
| 09:01 | <annevk> | hsivonen, something like </svg><div/>x |
| 09:01 | <annevk> | ? |
| 09:02 | <annevk> | well, plus some content before it |
| 09:03 | <zcorpan_> | i take it there's already a test like "<svg></svg>"? |
| 09:04 | <hsivonen> | zcorpan_: AFAICT, it depends on EOF handling having or not having certain other fixes |
| 09:04 | <hsivonen> | but let's try it |
| 09:04 | <hsivonen> | zcorpan_: <svg></svg> doesn't produce the wrong tree without the fix |
| 09:05 | <hsivonen> | given the current state of V.nu EOF handling |
| 09:06 | <zcorpan_> | the bug says that it can cause a crash, so seems good to include it |
| 09:07 | <hsivonen> | zcorpan_: already included |
| 09:07 | <zcorpan_> | ok |
| 09:07 | <hsivonen> | I need to step through this in debugger to understand why there's no bug... |
| 09:14 | <hsivonen> | so "A start tag, if the current node is an element in the HTML namespace." handles the normal start tag case |
| 09:15 | <hsivonen> | hmm. maybe I should try something more evil like <svg></svg><![CDATA[]]> |
| 09:15 | <hsivonen> | woohoo! a test case! |
| 09:17 | <zcorpan_> | that's the spirit! think evil :) |
| 09:46 | <hsivonen> | malignmark is my favorite MathML element name |
| 09:47 | <MikeSmith> | heh |
| 09:47 | <jgraham> | hsivonen: I sonsidered "don't crash" to be a good TC for that change :) |
| 09:48 | <jgraham> | But the CDATA thing is neat |
| 09:49 | <othermaciej> | hsivonen: doesn't that tend to piss him off? |
| 09:49 | <jgraham> | Also: malignmark is indeed quite awesome |
| 09:50 | <Workshiva> | Could he sue w3c for that kind of harassment? |
| 09:53 | <hsivonen> | Isn't the "still in foreign content" check redundant in http://html5.org/tools/web-apps-tracker?from=5299&to=5300 ? |
| 09:53 | <hsivonen> | how could it not still be in foreign at that point? |
| 09:56 | <Hixie> | yeah that does seem redundant |
| 09:56 | <Hixie> | since you have to be in foreign to get to that clause, and popping can't change the mode |
| 09:56 | <Hixie> | right? |
| 09:56 | <ashaw> | with so many web browsers that are colour corrected, I was wondering if others thought it would be useful to be able to set the input colour correction profile in HTML5? |
| 09:56 | <hsivonen> | Hixie: right. I'll file a bug. |
| 09:57 | <Hixie> | ashaw: you could, using @color-profile http://www.w3.org/TR/2003/CR-css3-color-20030514/#atcolorprofile but it was dropped because nobody implemented it |
| 09:58 | <ashaw> | do you know why not? |
| 09:59 | <zcorpan_> | annevk: you're listed in the acks section in dom core :) |
| 09:59 | <Hixie> | probably nobody cared, but i have no direct information |
| 09:59 | <Hixie> | annevk and TabAtkins might know |
| 10:04 | <hsivonen> | I wonder if rev 5300 causes an infinite loop on EOF |
| 10:04 | hsivonen | looks closer |
| 10:05 | <hsivonen> | I think there's an infinite loop |
| 10:06 | <ashaw> | also the css3 color-profile rule states that the default profile should be sRGB, this is not implemented in Safari, is this a bug? |
| 10:08 | <hsivonen> | yep, there definitely is an infinite loop |
| 10:08 | <hsivonen> | time to file a spec bug... |
| 10:08 | <jgraham> | hsivonen: where? |
| 10:08 | <hsivonen> | jgraham: <math><mi>EOF keeps reprocessing the EOF token |
| 10:09 | <zcorpan_> | clearly getting foreign lands right is hard |
| 10:12 | <jgraham> | Hmm, I thought I hd thought about that case at the time |
| 10:12 | <jgraham> | I have a feeling that I might have been misled by not actually reprocessing the token, since that just affects parse errors |
| 10:13 | <myakura> | ashaw: iirc they've talked something about color correction stuff last year though i'm not sure if anything's happend since then. |
| 10:17 | <hsivonen> | jgraham: I think getting more than one parse error in that case provides no value |
| 10:17 | <hsivonen> | certainly no value that'd justify complexity |
| 10:18 | <jgraham> | Yes, I think when the psec has the choice between simplicity and theoretical purity of #parse errors, spimplicity wins every time |
| 10:19 | <jgraham> | Ubuntu is really very unresponsive |
| 10:19 | <jgraham> | A multisecond delay on everything I type is bad |
| 10:19 | <hsivonen> | I wonder if a security update has broken it some time during August |
| 10:20 | <hsivonen> | since I returned from vacation, compiling Firefox has started making the mouse cursor and Spotify really jumpy |
| 10:20 | <hsivonen> | with 8 hardware threads |
| 10:21 | <jgraham> | Similar problem here |
| 10:21 | <jgraham> | Fewer threads but a less impressive machine |
| 10:22 | <jgraham> | and not Firefox :) |
| 10:27 | <annevk> | zcorpan_, I haven't touched editors/acknowledgments yet |
| 10:27 | <annevk> | zcorpan_, maybe I should |
| 10:32 | <annevk> | ashaw, it's a more complicated feature basically |
| 10:32 | <annevk> | ashaw, it will probably return at some point |
| 10:32 | <ashaw> | why is it more complicated> |
| 10:32 | <ashaw> | ? |
| 10:33 | <annevk> | yay, 8MiB per second; finally a somewhat reasonable internet connection |
| 10:33 | <annevk> | ashaw, it's more complicated than the other CSS color features |
| 10:42 | <erlehmann> | annevk, and i am on GRPS ;_; |
| 10:43 | <annevk> | poor you |
| 10:43 | <annevk> | everyone should have at least 100mbit |
| 10:43 | <erlehmann> | i second that motion |
| 10:44 | <ashaw> | I am in Australia, until last year I had 512Kbps adsl, now somewhat 8mbit/s ADSL2+ cannot wait until the NBN comes to where I am. |
| 10:45 | <annevk> | whoa |
| 10:45 | <ashaw> | yeah, |
| 10:45 | <ashaw> | looong copper lines. |
| 10:45 | <Lachy> | ashaw, I guess you must be in one of the lucky regional areas that will get prioritised over the next 3 years |
| 10:45 | <ashaw> | no |
| 10:45 | <ashaw> | newcastle. |
| 10:45 | <ashaw> | those areas can barely get 3mbit/s |
| 10:46 | <ashaw> | or less |
| 10:46 | <Lachy> | wow. I didn't expect Newcastle to have such bad ADSL at this stage. |
| 10:46 | <ashaw> | the exchange is 1km away/ |
| 10:46 | <ashaw> | actually lake macqurie |
| 10:46 | <Lachy> | hmm. that's strange. The copper lines must be really bad quality to have signal that low |
| 10:47 | <ashaw> | waterlogged. |
| 10:47 | <ashaw> | and no not really |
| 10:47 | <ashaw> | if you look at the graphs |
| 10:48 | <ashaw> | at the moment I am gettinh 9108000 baud |
| 10:48 | <ashaw> | and only 29.0dB attenuation |
| 10:48 | <ashaw> | which is pretty good. |
| 10:48 | <ashaw> | ADSL is just not as good as you think |
| 10:49 | <Lachy> | but surely if you're really getting ADSL2+, you'd be able to get more than 8Mbps. Are you sure you're not on ADSL1? |
| 10:49 | <ashaw> | no. |
| 10:49 | <ashaw> | it is ADSL2+ annex M, or so my modem says. |
| 10:50 | <Lachy> | ok |
| 10:50 | <ashaw> | the speed falls in a curve. |
| 10:52 | <ashaw> | anyway what sort of speed are you getitng. |
| 10:52 | <Lachy> | http://www.internode.on.net/residential/broadband/adsl/extreme/performance/ |
| 10:52 | <ashaw> | yes I know. |
| 10:52 | <Lachy> | according to that graph, at only 1km from the exchange, you should be getting up around 20 |
| 10:53 | <ashaw> | yeah, but I am not. my line looks electrically like 2.3km |
| 10:53 | <Lachy> | I'm overseas right now. But when I was in Aus in Northern Sydney, I believe I managed to get about 13Mbps on ADSL2+. But that was about 4 years ago. |
| 10:54 | <ashaw> | yeah, where are you now? |
| 10:54 | <Lachy> | Oslo |
| 10:54 | <ashaw> | hahaha. |
| 10:54 | <ashaw> | what project do you work on? |
| 10:54 | <ashaw> | to be here? |
| 10:55 | <Lachy> | I'm on Cable at home here. I get about 17Mbps peak. I could get up to 50Mbps if I could afford it. |
| 10:55 | <Lachy> | I work at Opera |
| 10:55 | <ashaw> | how much? |
| 10:55 | <ashaw> | ah. |
| 10:56 | <Lachy> | on my current plan, it works out to roughly $90/month depending on exchange rate |
| 10:56 | <ashaw> | download limit? |
| 10:56 | <Lachy> | but Norway is a very pricey country, so not bad overall |
| 10:56 | <Lachy> | haha. Usage caps don't exist in Europe :-) |
| 10:57 | <ashaw> | really? |
| 10:57 | <Lachy> | yes, really. |
| 10:57 | <Lachy> | at least, not in Norway, and I don't believe they do in any other country. But I could be wrong. |
| 10:57 | <ashaw> | I cannot wait for that to come here, as it will. |
| 10:57 | <ashaw> | Internode just released a 1TB plan. |
| 10:57 | <Lachy> | AAPT has unlimited |
| 10:58 | <Lachy> | iiNet has 1TB too |
| 10:58 | <Lachy> | (500GB on/off peak) |
| 10:59 | <ashaw> | I HATE AAPT. |
| 10:59 | <ashaw> | iinet is good though. |
| 11:00 | <Lachy> | I expect with the NBN, with reasonable wholesale prices and with the available bandwidth being more than enough to handle everyone, bandwidth caps will be a thing of the past |
| 11:00 | <Lachy> | I have no experience with AAPT |
| 11:00 | <Lachy> | iiNet were good for me though |
| 11:00 | <ashaw> | major problem is actually international trunks. |
| 11:00 | <Philip`> | Many UK ISPs seem to have usage caps |
| 11:01 | <Philip`> | like 10GB for the cheapest options and 40GB for slightly less cheap ones |
| 11:01 | <ashaw> | UK is very expensive though. |
| 11:01 | <Lachy> | woah. That's worse than Australia. WTF? |
| 11:01 | <ashaw> | yeah |
| 11:01 | <ashaw> | ] |
| 11:05 | <ashaw> | before, the main problem with colour calibration in browsers was Flash, is that still so? |
| 11:07 | <annevk> | no caps in the Netherlands |
| 11:11 | <annevk> | http://twitter.com/Rahul/status/23994111511 -- CNN on HTML5: "[...] an open-source platform that updates animation features of programs like Adobe's Flash." http://goo.gl/qXLV |
| 11:12 | <ashaw> | hahaha. Wroooong |
| 11:29 | <Lachy> | annevk, I don't see where in the linked article that can be found, nor anything else about HTML5 |
| 11:30 | <Lachy> | annevk, also, thanks for leaving me the Kahlua yesterday |
| 11:33 | <annevk> | maybe they fixed it? |
| 11:33 | <annevk> | oh well |
| 11:42 | <Lachy> | http://www.macrumors.com/2010/09/09/vlc-video-player-coming-to-ipad/ |
| 11:44 | <Lachy> | I assume that means they'll bring WebM and Theora support to the iPad, at least to the extent possible. |
| 11:51 | <DefV> | yeah, but it won't be hardware-accelerated like with x264 :-( |
| 11:51 | <ashaw> | why not? |
| 11:51 | <ashaw> | I mean why could it not be? |
| 11:52 | <gsnedders> | I expect the DSP is locked |
| 11:52 | <ashaw> | ewww. |
| 11:53 | <gsnedders> | It is on most phones, and I expect it is on the iPad |
| 12:13 | <hsivonen> | not a long time ago VLC folks claimed Apple doesn't them on Mac |
| 12:13 | <hsivonen> | if true, I wonder how the AppStore approval will go |
| 12:18 | <zcorpan_> | hsivonen: s//want/ ? |
| 12:18 | <hsivonen> | zcorpan_: yes |
| 12:26 | <hsivonen> | whoa, whoa. when has the meta prescan started to require http-equiv="content-type" for the content="..." case? |
| 12:29 | <zcorpan_> | http://www.w3.org/Bugs/Public/show_bug.cgi?id=9225 ? |
| 12:30 | <hsivonen> | zcorpan_: thanks |
| 12:43 | <hsivonen> | does anyone happen to have data about browser support of <meta http-equiv="pragma" value="no-cache"> ? |
| 12:51 | <hsivonen> | why doesn't this get cached? http://hsivonen.iki.fi/test/moz/pragma-no-cache-baseline.php |
| 12:55 | <zcorpan_> | maybe because it doesn't have an expires header? |
| 12:55 | <hsivonen> | hmm. should the baseline have an Expires header? |
| 12:55 | <hsivonen> | ah |
| 12:57 | <hsivonen> | now I need the right php format string for the expires header... |
| 12:57 | <hsivonen> | "r" |
| 12:57 | <hsivonen> | cool |
| 12:58 | <boogyman> | hsivonen: ##php should be able to help you out |
| 13:01 | <hsivonen> | boogyman: the manual was enough :-) |
| 13:11 | Philip` | finds it strange when there's spec-related discussions about references to garbage collected objects |
| 13:11 | <Philip`> | since surely if there's a reference to the object then it's not garbage, by definition |
| 13:13 | <zcorpan_> | discussion where? |
| 13:13 | <hsivonen> | I need to do s/value/content/ in the test case |
| 13:15 | <Philip`> | zcorpan_: On the WHATWG list recently, and in the spec itself in various places |
| 13:15 | <Philip`> | Seems like it's more about manual memory management than about GC |
| 13:18 | <hsivonen> | doh. I forgot to test IE8 in the quirks mode |
| 13:19 | <zcorpan_> | Philip`: like websockets with event listeners? |
| 13:20 | <hsivonen> | aaargh. IE supports the pragma stuff in the IE 5.5 mode |
| 13:20 | <hsivonen> | yay for IE modes to make developers of other browsers test more cases |
| 13:20 | <hsivonen> | s/to make/for making/ |
| 13:22 | <Philip`> | zcorpan_: Yes |
| 13:24 | <Philip`> | I think the usual definition is that a value is garbage after the last time it is read (and a conversative approximation is that it's definitely garbage after the last reference to it has been removed) |
| 13:24 | <Philip`> | so you can never tell the difference between a program that frees all garbage immediately and one that never frees garbage |
| 13:25 | <zcorpan_> | Philip`: isn't that the case with the spec also? |
| 13:26 | <Philip`> | The spec seems to care about the difference, else it wouldn't have to say anything about GC, so I assume it's got some undefined implementation-level notion of an imperfect garbage collection algorithm based on reference-counting and weak pointers or something |
| 13:29 | <zcorpan_> | if you create a websocket object and add an event listener to it, and remove the js-level reference to it, does the ua still have a reference to it because it has an event listener? |
| 13:32 | <Philip`> | If it is possible to observe a difference between an implementation that holds a reference and one that doesn't, then it must be because the potentially-collected value was still accessed somehow, which means it wasn't garbage, which means it shouldn't have been garbage collected |
| 13:32 | <zcorpan_> | yeah |
| 13:33 | <zcorpan_> | the spec just spells out that it is possible to observe a difference and spells out when it is no longer possible to observe a difference |
| 13:37 | <Philip`> | It spells it out in what seems like a peculiar level of abstraction, since it's talking about the mechanics of a garbage collection algorithm rather than about observable differences |
| 13:38 | <zcorpan_> | file a bug? |
| 13:39 | <hsivonen> | whoa. Opera doesn't cache http://hsivonen.iki.fi/test/moz/pragma-no-cache-quirks.php |
| 13:40 | <hsivonen> | so Opera has a quirks mode difference here like IE |
| 13:40 | <hsivonen> | sadness |
| 13:43 | <zcorpan_> | ouch |
| 13:46 | <hsivonen> | so the pragma is observed in Firefox 3.6, Opera quirks and IE quirks |
| 13:47 | <hsivonen> | and not observed in IE standards, Opera standards and WebKit |
| 13:47 | <hsivonen> | so currently WebKit is the only one not to observe the pragma in quirks... |
| 15:31 | <annevk> | DocumentType makes insertBefore insane |
| 15:32 | <annevk> | we should just make adoptNode support DocumentType and get rid of all the special casing |
| 15:41 | <hsivonen> | I'm starting to think the implicit closing of <p> might be *the* worst characteristic of HTML parsing |
| 15:42 | <hsivonen> | (from author POV) |
| 15:46 | <abarth> | hsivonen: it's really confusing to author |
| 15:46 | <abarth> | because it tricks you into thinking HTML works differently than it does |
| 15:47 | <zcorpan_> | hsivonen: why wouldn't you put the figure before the <p>? |
| 15:50 | <hsivonen> | zcorpan_: I want the top edge of the figure to align with the top edge of the paragraph text if the paragraph has a top margin or even padding plus border |
| 15:52 | <zcorpan_> | hsivonen: i take it you want figure to be phrasing content with transparent content model (plus figcaption)? |
| 15:52 | <hsivonen> | zcorpan_: it want it to be phrasing, yes |
| 15:53 | <hsivonen> | zcorpan_: I'd prefer non-transparent and scoping but maybe that's too much to ask for considering Degrade Gracefully |
| 15:53 | <zcorpan_> | hsivonen: what if you want it to align with the paragraph text but the figcaption contains several paragraphs? |
| 15:53 | <hsivonen> | zcorpan_: hence I'd prefer non-transparent and scoping |
| 15:54 | <zcorpan_> | maybe you should say that in the bug |
| 15:55 | <zcorpan_> | <figcaption> can still close <p> if <figure> is scoping |
| 15:57 | <zcorpan_> | although you also have this problem if you want a <table> to align with a paragraph, or a number of other things which also close <p> |
| 15:57 | <zcorpan_> | so i'm not convinced we should change this |
| 15:58 | <hsivonen> | zcorpan_: I think making <table> close <p> was a mistake |
| 15:58 | <hsivonen> | but Hixie put that one in Acid2, so water under the bridge |
| 15:58 | <hsivonen> | but that doesn't mean <figure> has to suck, too |
| 15:58 | <annevk> | hmm, insertBefore for DocumentFragment is prolly also wrong |
| 15:59 | <annevk> | as using insertBefore to "implement it would remove the nodes from the DocumentFragment |
| 15:59 | <annevk> | gah |
| 15:59 | <annevk> | these functions have lots of permutations, it's crazy |
| 16:10 | <gsnedders> | Hmm, I'm getting: "o: command not found", and I have gcc/g++ installed. WTF? |
| 16:11 | <hsivonen> | I learned something new today: http://www.utoronto.ca/web/HTMLdocs/NewHTML/multicol.html |
| 16:11 | <Philip`> | gsnedders: What command are you trying to run? |
| 16:12 | <hsivonen> | also, I was unaware that WebTV had minted multiple browser-specific elements that no one else supports |
| 16:12 | <gsnedders> | Philip`: Just trying to build mozilla-central tip (gah, really, no builds with --enable-tests? I think this everytime I need them…) |
| 16:14 | <jgraham> | I thought they had those builds now |
| 16:15 | <hsivonen> | <spacer> is about to become the longest-lived major engine-suppored HTML element to be eradicated |
| 16:15 | <Philip`> | Maybe the .tests.zip in http://ftp.mozilla.org/pub/mozilla.org/firefox/nightly/latest-trunk/ ? |
| 16:15 | Philip` | has no idea what they are, but they do say "tests" in them |
| 16:16 | <hsivonen> | <layer> and <ilayer> lasted only for one major release. <multicol> two. <hype> was Mac-only. |
| 16:16 | <hsivonen> | has Opera minted any elements? |
| 16:16 | <hsivonen> | WebKit has minted <canvas> but that one became standard |
| 16:16 | gsnedders | still wants to know why it won't compile |
| 16:17 | <gsnedders> | hsivonen: In HTML? Not that we currently support, at least |
| 16:17 | <hsivonen> | gsnedders: ok. Wikipedia doesn't know of any, either. |
| 16:18 | <hsivonen> | I think <spacer> is also about to become the first HTML element to be unsupported by Gecko after having been supported by Gecko |
| 16:19 | <gsnedders> | I know we dropped support for the bq element :P |
| 16:20 | <hsivonen> | gsnedders: Wikipedia is unaware! |
| 16:20 | <hsivonen> | I wonder if the IRC log works as an authoritative source for Wikipedia... |
| 16:21 | <gsnedders> | heh |
| 16:21 | <gsnedders> | Basically, we used HTMLQuoteElement for bq |
| 16:21 | <gsnedders> | That was the extent of our support for it. |
| 16:29 | <hsivonen> | jgraham: we have no way to represent a carriage return in expected tree builder output, right? |
| 16:30 | <hsivonen> | the test with FOO
ZOO is the input is bogus. I'm wondering what to do about it |
| 16:30 | <gsnedders> | Why would a raw 0x0D byte now work? |
| 16:31 | <hsivonen> | gsnedders: wouldn't text editors break it upon next edit? |
| 16:31 | <hsivonen> | gsnedders: and it doesn't work when the test data is loaded via iframe |
| 16:31 | <jgraham> | That could be a problem |
| 16:31 | <jgraham> | Not sure what to do about it though |
| 16:31 | <hsivonen> | (maybe I should change our harness to use XHR but I'd rather not, since everything else works via iframe) |
| 16:32 | <hsivonen> | I could remove the test and make sure the tokenizer tests cover 
 |
| 16:32 | <jgraham> | hsivonen: Why does the iframe present a problem? |
| 16:33 | <jgraham> | Does some normalisation happen if you load a data: uri? |
| 16:33 | <hsivonen> | jgraham: if you load text/plain in a browsing context, line break normalization happens |
| 16:34 | <jgraham> | Oh text/plain |
| 16:35 | <jgraham> | In fact it seems generally problematic |
| 16:35 | <hsivonen> | the use of an iframe might be caused by the harness been developed to work with file: URLs |
| 16:35 | <hsivonen> | dunno. sayrer would know |
| 16:35 | <jgraham> | Unless you keep the expected output e.g. percent encoded |
| 16:36 | <hsivonen> | jgraham: the files that get loaded as text/plain;charset=utf-8 in the iframes are the html5lib .dat files verbatim |
| 16:36 | <jgraham> | hsivonen: Yeah, it is not easy with your approach |
| 16:37 | <gsnedders> | http://pastebin.com/TgGNx2FF — fairly weird. |
| 16:38 | <gsnedders> | I guess I should try asking in a proper channel… |
| 16:40 | <hsivonen> | for the tests of the nature <html>�<frameset></frameset> I haven't yet checked with a hex editor if the test is bogus or if the harness is |
| 16:41 | <hsivonen> | the tests are still indicating 5 Gecko/V.nu bugs |
| 16:42 | <hsivonen> | and 6 if the above REPLACEMENT CHARACTER thing isn't a harness/test bug |
| 16:46 | <gsnedders> | I guess libvpx is trying to do something different to the rest of the moz build systems |
| 16:50 | <annevk> | jgraham, how easy would it be to let testharness not quit at the first non-fatal failure? |
| 16:50 | <hsivonen> | gsnedders: do you have yasm installed? |
| 16:51 | <jgraham> | annevk:All failures are fatal |
| 16:51 | <jgraham> | Unless they are in seperate tests |
| 16:51 | <Philip`> | gsnedders: Looks kind of like it's trying to execute half a line (I'd expect it to be "something ... -o emms.o -f elf64 ...") |
| 16:51 | <jgraham> | It's a pretty fundamental part of the design |
| 16:51 | <annevk> | i guess i could use nested tests more often |
| 16:51 | <annevk> | okay |
| 16:52 | <jgraham> | You don't need to nest |
| 16:52 | <jgraham> | Just have multiple test(function(){}) per file |
| 16:52 | <annevk> | yeah okay |
| 16:52 | <gsnedders> | hsivonen: yes |
| 16:53 | <gsnedders> | Philip`: I got that far :P |
| 16:53 | <gsnedders> | Running it with no parallelizisim appears to be working |
| 16:55 | <annevk> | insertBefore/appendChild/replaceChild are some of the most complex methods in whole of DOM Core |
| 16:55 | <annevk> | though compareDocumentPosition is pretty bad too |
| 17:52 | <jgraham> | I hate it when things sell for < 300 USD yet even in sales are 5000 SEK |
| 17:53 | <jgraham> | That is _more_than_a_factor_or_two_ |
| 17:56 | <annevk> | that's the price of freedom |
| 18:12 | <annevk> | i wonder how many steps it would be to spell out the HIERARCHY_REQUEST_ERR cases |
| 18:13 | <annevk> | the easy solution would of course be "do not violate the nodes model" but that is very easy |
| 18:14 | <annevk> | but just considering the amount of complexity you get for DOCUMENT_TYPE or the root ELEMENT I'm not sure I want to go there |
| 18:14 | <annevk> | hmm |
| 18:19 | <annevk> | // test 25: test namespace checking in createDocumentType |
| 18:19 | <annevk> | boo |
| 18:20 | <annevk> | Acid3 tests a whole bunch of crappy DOM stuff |
| 18:21 | <sicking> | s/DOM// |
| 18:22 | <annevk> | heh, true |
| 18:22 | <annevk> | oh well, we're not gonna repeat that mistake for Acid4 |
| 18:24 | <sicking> | annevk: riiiiiight |
| 18:25 | <sicking> | also, i think we need to make changes to acid3 |
| 18:25 | <sicking> | it's really stupid if we're adding things to the web platform just to make hixie happy |
| 18:26 | <sicking> | or rather, one page that hixie wrote happy |
| 18:26 | <annevk> | I don't think he's happy with it |
| 18:26 | <sicking> | right, but it's still about keeping that one page from breaking |
| 18:27 | <annevk> | yeah, I'm not at all opposed to changes to Acid tests |
| 18:27 | <annevk> | you're right actually, maybe we actively should make changes |
| 18:28 | <annevk> | not changing because we thought "something was set in stone" has made things bad in the past, too |
| 18:28 | <annevk> | e.g. CSS margin collapsing |
| 19:40 | <hsivonen> | Acid3 will bite us when we do XML5 |
| 19:47 | <othermaciej> | does Acid3 test for XML strictness? |
| 19:49 | <hsivonen> | othermaciej: it does |
| 19:49 | <hsivonen> | character encoding strictness at least |
| 19:54 | <Hixie> | acid3 tests a bunch of stuff that the specs required that the browsers didn't do, which we have since largely come to view as suboptimal |
| 19:54 | <Hixie> | it's sad |
| 19:55 | <Hixie> | acid2, and to some extent acid3, both predate the time where i was comfortable with just making wholesale changes to long-established specs |
| 19:55 | <Hixie> | (acid3 does sidestep some problems, like making Attr nodes optional) |
| 19:56 | <Hixie> | anyway, if the specs change then i'll happily change acid3 |
| 20:14 | gsnedders | blatantly has bad Mozilla karma today. Thunderbird appears to have lost all email. |
| 20:18 | <gsnedders> | All my account settings are there, but the list of folders is empty, and the list of emails doesn't even appear because there's no folder selected (because there are no folders) |
| 20:18 | <gsnedders> | What on earth? |
| 20:23 | <gsnedders> | http://stuff.gsnedders.com/thunderbird.png — awesome! |
| 20:30 | <gsnedders> | folderTree.json corrupt. Nice. |
| 20:44 | <ap> | Hixie: is there a spec that does/could specify what the following does? http://nypop.com/~ap/webkit/awstest.html |
| 20:44 | <ap> | WebKit matches IE, but not Firefox |
| 20:46 | <TabAtkins> | ap: What exactly is being tested here? The use of try/catch in on* attributes? Or naming a function the same as a @name in the document? Something else? |
| 20:46 | <zcorpan_> | ap: the question is whether <elm id=foo> overrides function foo() {} ? |
| 20:46 | <ap> | TabAtkins: the latter (i.e. order of lookup on window object) |
| 20:46 | <ap> | zcorpan_: name or id, whatever |
| 20:48 | <ap> | zcorpan_: the relative precedence of id and name would be HTML5 domain, I think |
| 20:48 | <zcorpan_> | http://www.whatwg.org/specs/web-apps/current-work/multipage/browsers.html#named-access-on-the-window-object |
| 20:48 | <zcorpan_> | "supports named properties" is a webidl term iirc |
| 20:49 | <zcorpan_> | http://www.w3.org/TR/WebIDL/#dfn-support-named-properties |
| 20:49 | <ap> | zcorpan_: I'm not sure if that answers my question (I think not) |
| 20:50 | <ap> | zcorpan_: HTML5 talks about DOM, naturally, not about JS functions |
| 20:52 | <jgraham> | Would you not expect it to depend on the order in which they are defined |
| 20:52 | <zcorpan_> | ap: i would expect webidl to answer this question in its JS binding section, although i don't know if it does currently |
| 20:54 | <Workshiva> | You probably don |
| 20:54 | <Workshiva> | don't want to link the WebIDL from 2008 |
| 21:02 | <zcorpan_> | oops |
| 21:03 | <zcorpan_> | google usually gives me the dev.w3.org version |
| 21:03 | <Workshiva> | It seems to have stopped |
| 21:03 | <Workshiva> | I remember accidentally reading the TR version a while ago |
| 21:06 | <zcorpan_> | horrifying experience? |
| 21:07 | <Workshiva> | I couldn't find the property I was looking for (because it hadn't been created in 2008) |
| 21:15 | <jgraham> | /TR/ is evil |
| 21:19 | <jgraham> | Also, why are the hybi people discussing compression? We don't even have a functional protocol yet |
| 21:20 | <jgraham> | and the four weeks timeline is long past |
| 21:20 | <TabAtkins> | Because they have no clue what they're doing? |
| 22:06 | <zcorpan_> | <http://www.w3.org/mid/9B2DE9094C827E44988F5ADAA6A2C5DACD486B⊙Hppc>; |
| 22:07 | <zcorpan_> | W3C is pleased to announce the advancement of "Associating Style |
| 22:07 | <zcorpan_> | Sheets with XML documents 1.0 (Second Edition)" to Proposed Edited |
| 22:07 | <zcorpan_> | Recommendation: |
| 22:07 | <zcorpan_> | http://www.w3.org/TR/2010/PER-xml-stylesheet-20100909/ |
| 22:10 | <zcorpan_> | seems like he switched back to using the xhtml-generating version of the xslt sheet. oh well |
| 22:46 | <Hixie> | ap: WebIDL (and the IDL block and prose in the HTML5 spec) define it |
| 23:05 | <TabAtkins> | Is it appropriate for me to remove a TrackerRequest for a bug being escalated inappropriately? |
| 23:05 | <TabAtkins> | Or is that something only editor/chairs should do? |
| 23:07 | <Hixie> | not something for the editor to do |
| 23:07 | <Hixie> | that would be a conflict of interest |
| 23:08 | <TabAtkins> | K. Then, othermaciej, can I kill TrackerRequest from 10465, or should I let you do so? Faulkner is attempting to escalate a bug because of a different bug. |
| 23:09 | <TabAtkins> | (It's not even just trying to fit two bugs in one - there's actually another existing bug *from him* that actually covers the issue he's trying to escale 10465 over.) |
| 23:28 | <othermaciej> | TabAtkins: it does seem like it would be better for Steve to escalate bug 10478 if he intends to, though that bug in my opinion needs a PFWG response |
| 23:29 | <TabAtkins> | Oh, I agree with that. The issue at hand, though, is just the handling of 10465, which is completely wrong. |
