| 00:18 | Hixie | sends another private e-mail to kristof trying to convince him to use proper quoting style |
| 01:32 | <Hixie> | i'm down to only 107 e-mails from henri to deal with |
| 01:33 | <Hixie> | and there are 1500 or so e-mails in folders i've deferred for one reason or another, and only 500 others |
| 01:33 | <Hixie> | the end is in sight! |
| 01:33 | <Hixie> | only a year or two to go |
| 01:33 | <Hixie> | to get this feedback to sane numbers |
| 01:34 | <csarven> | You should just text-to-speech and listen to all the emails while you sleep! :P |
| 01:34 | <Hixie> | i've read most of them already |
| 01:49 | <Hixie> | why does firefox2 scroll the page when the notification widget comes up |
| 01:49 | <Hixie> | weird |
| 01:49 | <Hixie> | i wish i could use ff3 on my linux box |
| 01:58 | <BlueG> | I am curious about the relationship between the WHATWG and the W3C HTML WG. Ian Hickson is in both groups and is the editor for the spec, right? |
| 01:58 | <Hixie> | i am, yes |
| 01:59 | <Hixie> | html5 is sort of a joint venture between the two groups |
| 01:59 | <BlueG> | ok, I am not entire sure I understand why there are two working groups and what is the difference in their roles? |
| 01:59 | <Hixie> | it's sort of a historical thing |
| 01:59 | <BlueG> | ok |
| 01:59 | <Hixie> | whatwg started off in 2004 or so when the w3c told the browser vendors who founded the whatwg that the w3c didn't want to work on html5 |
| 01:59 | <Hixie> | they then founded the whatwg to work on html5 in public |
| 02:00 | <BlueG> | so, people who are interested in the developing spec can join either, or both wg? |
| 02:00 | <Hixie> | later on (late 2006ish to early 2007) the w3c changed their minds and asked if the html5 work could happen in the w3c, so it now happens in both |
| 02:00 | <Hixie> | yup |
| 02:01 | <Hixie> | http://blog.whatwg.org/w3c-restarts-html-effort has instructions on how to join the w3c group |
| 02:01 | <BlueG> | ok |
| 02:01 | <Hixie> | http://www.whatwg.org/mailing-list#specs has instructions on joining the whatwg group |
| 02:02 | <BlueG> | what are the current differences between the two? |
| 02:02 | <Hixie> | the main difference is that in theory microsoft participate only in the w3c list |
| 02:02 | <Hixie> | in practice they don't really participate there either very much |
| 02:03 | <BlueG> | ok |
| 02:05 | <Hixie> | there are certain topics that just get discussed more on one list than the other |
| 02:06 | <Hixie> | for example video and the new web workers stuff is more discussed on the whatwg list |
| 02:06 | <Hixie> | and <img alt=""> and <table> semantics are more discussed on the w3c list |
| 02:06 | <BlueG> | so it is probably good to join both, at least if you have broad interest in the development |
| 02:06 | <Hixie> | but that's mostly a function of which list people who care about those topics are primarily on |
| 02:06 | <Hixie> | yeah, though to be honest you'll find that's a lot of traffic :-) |
| 02:06 | <BlueG> | ok |
| 02:07 | <Hixie> | it's died down now a bit, and the w3c list gets about 500 e-mails a month, but there was a time where 1200 e-mails amonth wasn't uncommon |
| 02:07 | <Hixie> | whatwg has a lot less traffic, and it tends to be somewhat more technically focussed |
| 02:08 | <Hixie> | but that's mostly because the whatwg list is an older and more mature community now, so expected behaviour is more understood |
| 02:09 | <Hixie> | (whatwg has about 300 a month) |
| 02:12 | <BlueG> | it looks easier to join whatwg |
| 02:12 | <Hixie> | there's that too :-) |
| 02:12 | <BlueG> | haha |
| 02:13 | <BlueG> | ok |
| 02:14 | <Hixie> | you can also read the w3c list online if you don't want to post much: http://lists.w3.org/Archives/Public/public-html/latest |
| 02:14 | <BlueG> | ok |
| 02:16 | <Hixie> | this irc channel is sort of a hub for a lot of the html5 activity (mostly because i rant on here while i'm editing the spec :-) ) so if you hang out here you'll likely here about anything interesting going on |
| 02:17 | <Hixie> | there are logs for this channel too, see the /topic for details |
| 02:20 | <BlueG> | ok |
| 02:26 | <BlueG> | there seems to be a problem with the signup page for whatwg, its throwing an error |
| 02:26 | <Hixie> | hm, crap |
| 02:26 | Hixie | looks |
| 02:33 | <Hixie> | try sending mail to whatwg-join⊙wo no idea what's up with mailman. |
| 02:33 | <Hixie> | i'll have to send dreamhost a support request if it's still broken tonight |
| 02:35 | <BlueG> | ok |
| 02:36 | <Hixie> | right, dinner time. i'll be back in a bit. hopefully dev.w3.org's cvs server and whatwg's mailman server will be in a better state upon my return. |
| 04:09 | <BlueG> | whatwg-join⊙wo doesn't seem to have worked either, and the mailman page is still giving the same error |
| 05:40 | <Hixie> | BlueG: reported it, hopefully they'll fix it soon |
| 05:41 | <BlueG> | ok, thanks |
| 08:18 | <annevk> | hmm, wtf? http://forums.whatwg.org/viewtopic.php?t=201 |
| 08:27 | <hsivonen> | annevk: I don't see a usual spam payload. perhaps the author just has a lot to post. |
| 08:37 | <zcorpan> | Hixie: <script src=javascript:a=1> and <script src=javascript:"a=1"> are different |
| 08:37 | <zcorpan> | Hixie: i'm not sure i understand which you've specced but the first should work and the second not |
| 08:37 | <Hixie> | no, neither works |
| 08:38 | <annevk> | hsivonen, yeah... |
| 08:39 | <annevk> | Hixie, insertAdjacentHTML is at one point referred to as "attribute" |
| 08:39 | <zcorpan> | Hixie: oh. hmm. the first works in opera |
| 08:51 | <zcorpan> | annevk: thanks (re spam) |
| 09:31 | <zcorpan> | "It may have been a mere warning, but it made a lot, lot, lot of people anxious and upset. So, by popular demand – and also because the XHTML working group are preparing a revised note on XHTML and media types − the warning is gone." -- http://www.w3.org/QA/2008/08/markup_validator_updated.html |
| 09:31 | <Hixie> | zcorpan: yeah, zcorpan (i think) mentioned that already |
| 09:31 | <Hixie> | er |
| 09:31 | <Hixie> | s/zcorpan/annevk/ |
| 09:32 | <Hixie> | zcorpan: what was the waring? |
| 09:32 | <Hixie> | oh i see |
| 09:33 | <Hixie> | how silly |
| 09:33 | <Hixie> | oh well |
| 09:33 | <Hixie> | i look forward to henri's validator giving the w3 one a run for its money |
| 09:55 | hsivonen | notes that comments can be implemented with less buffer copying if stuff is always "appended" but emitting the comment tokens trims the right number of trailing hyphens |
| 09:56 | <hsivonen> | the joke is that the SGML people used the longest names for the most common things they talked about |
| 09:57 | <hsivonen> | but as takkaria abserved about the # of states for doctypes, it seems that the importance of a given piece of SGML-legacy syntax in inversely proportional to its usefulness |
| 09:58 | <hsivonen> | it's dumb how many states go for doctypes and comments compared to tags |
| 10:00 | <hsivonen> | I need to eliminate 129 bytes from my tokenizer loop |
| 10:14 | <Hixie> | hsivonen: the only reason tags are simpler is because we skipped out on 90% of the syntax for tags, whereas for doctypes we went all-out in doing the sgml-compatible thing |
| 10:17 | <hsivonen> | Hixie: what did you skip? <> and </> still invoke stuff even though different stuff |
| 10:17 | <hsivonen> | good thing we aren't doing internal subsets |
| 10:18 | <Hixie> | <foo/ /, <> and </> implying the right tags, etc |
| 10:18 | <hsivonen> | I should go read the XML archives and find out who perpetrated all the DTD stuff |
| 10:18 | <hsivonen> | oh right, <foo/ / |
| 10:18 | <Hixie> | philip taylor (w) claims that <script src=javascript:"alert(1)"></script> shows an alert in SeaMonkey 1.1.11 |
| 10:19 | <hsivonen> | I mean, one has to respect Goldfarb for starting all this, but there would have been some serious opportunity for syntax simplification |
| 10:20 | <annevk> | I think SeaMonkey is Gecko 1.8, not 1.9 |
| 10:22 | <Hixie> | yeah that was my guess too |
| 10:23 | <annevk> | he's probably right in that it's possible to change |
| 10:23 | <hsivonen> | 134 bytecodes to cut... |
| 10:23 | hsivonen | doesn't like arbitrary VM limits |
| 10:25 | <hsivonen> | Sun is now all like "there are more JVM languages than Java" |
| 10:25 | <hsivonen> | but then they keep around this silly 8000 byte limit that makes it harder to compile other languages into bytecode |
| 10:25 | <hsivonen> | or to write huge state machines by hand |
| 10:26 | <Hixie> | what is this limit? |
| 10:27 | <hsivonen> | Hixie: if a method takes more than 8000 bytes, it doesn't get JITed |
| 10:27 | <hsivonen> | which sucks *badly* |
| 10:27 | <Hixie> | that's odd |
| 10:31 | <Hixie> | so about 1% of files in philip's sample have meta content-language |
| 10:31 | <hsivonen> | it seems that the Eclipse Java Compiler outputs very naïve bytecode |
| 10:31 | <Hixie> | i suppose i should check what fraction of those have useful content-language values that aren't redundant with other language information |
| 10:31 | <hsivonen> | I guess I should check if javac already compiles this to under 8000 bytes |
| 10:31 | <hsivonen> | Hixie: FrontPage outputs Content-Language |
| 10:32 | <Hixie> | usefully? |
| 10:32 | <hsivonen> | and it affects spell checking, which is good for getting people to change it |
| 10:32 | <hsivonen> | Hixie: yes, unless it gets overriden by <span lang> and the user doesn't notice |
| 10:32 | <Hixie> | sigh |
| 10:33 | <hsivonen> | Hixie: they tried to be really smart and tied the input language to the input method |
| 10:33 | <Hixie> | with lang="", xml:lang="", meta content-language treated as lang, and maybe even http content-language treated per mime, determining the language of a node is going to be one heck of a pain in the ass |
| 10:33 | <hsivonen> | which works if your universe only includes American English, Japanese, Chinese, Korean and Hebrew |
| 10:34 | <hsivonen> | but it sucks pretty badly when Europeans use their national keyboards for English |
| 10:36 | <annevk> | Hixie, Firefox at least already does that for :lang() |
| 10:37 | <annevk> | Hixie, we have some bugs on Opera to get it to do the same (or it might already have been fixed, forgot) |
| 10:37 | <annevk> | Content-Language, <meta http-equiv=content-language>, lang="" |
| 10:38 | <Hixie> | so what happens when Content-Language has multiple languages set? |
| 10:39 | <annevk> | pick the first |
| 10:39 | <zcorpan> | i don't particularly like that content-language is involved at all |
| 10:39 | <zcorpan> | but oh well |
| 10:41 | <Hixie> | m either |
| 10:43 | <Hixie> | should i make the Content-Language pragma non-conforming? |
| 10:43 | <Hixie> | i guess there's no point |
| 10:44 | <annevk> | we could try to remove support for it from browsers I suppose |
| 10:46 | <hsivonen> | I've already written code that supports <meta http-equiv=content-language> between <html lang> and HTTP |
| 10:46 | <annevk> | and you'd be sad to remove it? :) |
| 10:47 | <hsivonen> | yeah :-) |
| 10:57 | <zcorpan> | our developers don't mind removing code :) |
| 10:58 | <zcorpan> | but we'd have to change :lang() in that case too, no point being inconsistent |
| 10:58 | <Hixie> | content-language pragma defined |
| 11:11 | <annevk> | zcorpan, theoretically :lang() would match the language of the element, in which case removing support for content-language would "just work" |
| 11:18 | <annevk> | Hixie, "This pragma not exactly equivalent to the HTTP" |
| 11:21 | <Hixie> | thx |
| 11:23 | <annevk> | Hixie, the first note in "Dynamic markup insertion" has that other error I was trying to get you to fix |
| 11:23 | <annevk> | Hixie, the relevant bit reads "The outerHTML and insertAdjacentHTML attributes, on the other hand, only apply to Element nodes." |
| 11:25 | <Hixie> | already fixed |
| 11:26 | <zcorpan> | Hixie: the spec doesn't ban multiple content-language declarations |
| 11:26 | <zcorpan> | Hixie: also perhaps it should require <html lang> to match |
| 11:27 | <annevk> | Hixie, also, that's the only place the method is mentioned without parenthesis |
| 11:27 | <zcorpan> | (if present) |
| 11:27 | <Hixie> | yes it does, "There must not be more than one meta element with any particular state in the document at a time." |
| 11:27 | <zcorpan> | oh |
| 11:27 | <Hixie> | annevk: hm, good point |
| 11:27 | <Hixie> | zcorpan: i think matching the lang="" is unnecessary, though i agree a warning might be useful there if henri wants to give one |
| 11:28 | <annevk> | (another inconsistency I noticed was document.write() versus just write(); most methods are defined without accessor in front of them in the rest of the spec) |
| 11:28 | <Hixie> | yeah, i use the full name for a handful of them because that's how most people refer to them |
| 11:28 | <Hixie> | window.open() is the same iirc |
| 11:28 | <annevk> | window.open() has both styles |
| 11:28 | <Hixie> | yeah |
| 11:29 | <annevk> | it's <dfn>open()</dfn> but <dfn>document.write()</dfn>, etc. |
| 11:29 | <annevk> | oh well |
| 11:29 | <annevk> | maybe I should care about that after CR :) |
| 11:29 | <Hixie> | indeed :-D |
| 11:30 | <Hixie> | ok bed time |
| 11:30 | <Hixie> | nn |
| 11:32 | hsivonen | wishes Java allowed goto |
| 11:33 | <hsivonen> | all this high-levelness interferes with writing state machines |
| 11:42 | <zcorpan> | Hixie: are you saying that IE6 has less than 1% of the browser install base? :) |
| 12:19 | zcorpan | notes that <html xmlns="http://www.w3.org/1999/xhtml"; lang="en" xml:lang="sv"> validates in v.nu under xhtml 1.0 |
| 12:21 | <hsivonen> | zcorpan: which parser? |
| 12:22 | <zcorpan> | hsivonen: xml |
| 12:22 | <hsivonen> | makes sense |
| 12:22 | <zcorpan> | i had expected it to complain about the language mismatch |
| 12:22 | <hsivonen> | so many legacy holes to plug |
| 12:25 | <hsivonen> | zcorpan: recorded: http://bugzilla.validator.nu/show_bug.cgi?id=286 |
| 12:25 | <hsivonen> | thanks |
| 12:59 | <hsivonen> | performance tuning is hard |
| 13:08 | <annevk> | argh, I thought my camera could handle CompactFlash but it can't... |
| 13:08 | <annevk> | anyone in need of a CF card? |
| 13:12 | <hsivonen> | lesson learned today: never try doing two performance "optimizations" without benchmarking in between |
| 13:15 | <hsivonen> | another lesson learned today: it's more efficient to copy data than to check if data needs to be copied |
| 13:17 | <jgraham> | annevk: What CF card? |
| 13:17 | jgraham | was thinking of getting one |
| 13:17 | <jgraham> | (more) |
| 13:18 | <jgraham> | hsivonen: That latter lesson seems rather situation-specific |
| 13:19 | <hsivonen> | jgraham: very likely, yes. |
| 13:19 | <hsivonen> | anyway, I implemented copy avoidance for names and comments |
| 13:19 | <hsivonen> | and attribute values without entity refs |
| 13:20 | <hsivonen> | so that copy happened late only if buffer boundary fell inside the value of of an attribute value had a & in it |
| 13:21 | <hsivonen> | turns out that checking for each character if mode is copying or copy avoidance sucked big time |
| 13:22 | <hsivonen> | so at least doing an 'if' per charecter sucked |
| 13:22 | <hsivonen> | in non-Java languages, function pointers could be used to swap the appending function or something |
| 13:23 | <annevk> | jgraham, SanDisk 16GB Extreme III |
| 13:25 | annevk | is off to buy a SDHC card |
| 13:25 | <jgraham> | annevk: Wow that's big. |
| 13:25 | <annevk> | it was the biggest I could find, but apparently it didn't work for my camera :/ |
| 13:25 | <jgraham> | I'm happy to buy it off you assuming it's not too expensive |
| 13:26 | <annevk> | I paid 145 EUR |
| 13:26 | <jgraham> | Wow, that is really quite expensive |
| 13:26 | <annevk> | is it? |
| 13:26 | <jgraham> | Isn't it cheaper to get more smaller cards? |
| 13:26 | <hsivonen> | 2 GB is like 12 euros |
| 13:27 | <jgraham> | annevk: I just mean in absolute terms |
| 13:27 | annevk | compared it with 2 8GB cards and that didn't matter |
| 13:28 | <annevk> | I wanted a big one so I didn't have to swap and could do HD movies and such |
| 13:28 | <annevk> | seems that SDHC only maxes at 8 GB though for a lot less |
| 13:28 | <jgraham> | I bought a not-quite-as-good 4Gb card for £15ish which is like 80 euros for 16Gb |
| 13:29 | <jgraham> | The advantage of more smaller cards is that losing one in't such a big deal |
| 13:31 | <annevk> | well, mine would basically never leave my camera :) |
| 13:31 | <annevk> | anyways, really going now |
| 14:13 | <annevk> | ok, that was expensive, I should see into returning it |
| 14:14 | <annevk> | just bought SDHC 8GB Ultra II for 32 EUR |
| 15:00 | <zcorpan> | http://software.hixie.ch/utilities/js/live-dom-viewer/?%3C!DOCTYPE%20html%3E%0D%0A%3Cstyle%3Eembed%2C%20object%2C%20iframe%20%7B%20width%3A123px%3B%20height%3A45px%3B%20%7D%3C%2Fstyle%3E%0D%0A%3Cscript%3E%0D%0Aonload%3Dfunction()%7B%0D%0Avar%20e%20%3D%20document.body.firstChild%3B%0D%0Ado%20%7B%0D%0Aw(e.nodeName%20%2B%20'%3A%20'%20%2B%20e.width%20%2B%20'%2C%20'%20%2B%20e.height)%0D%0A%7D%20while%20(e%20%3D%20e.nextSibling)%3B%0D%0A%7D%0D%0A%3C%2F |
| 15:00 | <zcorpan> | script%3E%0D%0A%3Cembed%3E%3Cobject%3E%3C%2Fobject%3E%3Ciframe%3E%3C%2Fiframe%3E%3Cimg%3E%3Cvideo%3E%3C%2Fvideo%3E |
| 15:00 | <zcorpan> | (uploaded) |
| 16:45 | <jgraham> | It is the intent of the spec that <video><source type=type1><source type=type2><object></object></video> won't display the <object> if the browser doesn't support either type1 or type2 videos, right? (where is this specified normatively. I could only obviously see a should-level requirement when I scanned the spec). Would displaying the fallback in this case be helpful for migration from non <video> browsers without requiring a js-api for detecting |
| 16:50 | <Lachy> | jgraham, IIRC, the fallback for <video> is only intended for browsers that don't support the element at all. The spec says "Content may be provided inside the video element. User agents should not show this content to the user" |
| 16:50 | <Lachy> | I'm not sure why it only says should instead of must, though |
| 16:52 | <jgraham> | Lachy: Yeah that was the requirement I found. The use case it doesn't address is the wikimedia one where you want to use <video>+ogg for firefox, want to use an applet+ogg for <video>-supporting Safari and can't transcode for some reason, possibly legal |
| 16:53 | <jgraham> | (so <video>+mp4 for Safari is not possible) |
| 16:53 | <jgraham> | (where mp4 is H.264 or whatever it's called) |
| 18:40 | zcorpan | reads http://www.w3.org/MarkUp/2008/ED-xhtmlmime-20080618/ |
| 18:41 | zcorpan | is amused that the abstract contains rfc2119 words |
| 18:45 | <zcorpan> | "Elements and attributes in those document types belong to the XHTML namespace" |
| 18:57 | <zcorpan> | "However, authors should be aware that such a document [served as 'application/xml'] may not always be processed as XHTML (e.g. hyperlinks may not be recognized), depending on user agents." |
| 18:59 | <zcorpan> | "Authors SHOULD explicitly identify the XHTML namespace through the namespace declaration when they serve an XHTML Family document as 'application/xml' to facilitate the chance for reliable processing." |
| 18:59 | <zcorpan> | well i guess i could quote the whole thing here, it's all amusing |
| 19:11 | <zcorpan> | appendix c becomes appendix a |
| 19:14 | <zcorpan> | "The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in RFC 2119 [RFC2119]." |
| 19:14 | <zcorpan> | later on: |
| 19:14 | <zcorpan> | "Like all of this document, this Appendix is informative. It contains no absolute requirements, and should NEVER be used as the basis for creating conformance nor validation rules of any sort. Period." |
| 19:17 | <zcorpan> | more consice A1: Rationale: They are rendered in some crappy mobile browsers and they'll trigger quirks mode in IE |
| 19:20 | <zcorpan> | A2: Rationale: Netscape Navigator 4 would think the slash is part of the tag name when there are no attributes. Also HTML UAs interpret </br> as <br>. |
| 19:22 | <zcorpan> | A3: Rationale: since HTML UAs just ignore the slash it's interpreted as a start tag. |
| 19:24 | <zcorpan> | A4 is just bogus |
| 19:25 | <zcorpan> | A5 seems partially bogus |
| 19:26 | <zcorpan> | A7 doesn't give rationale for the DO NOT |
| 19:27 | <zcorpan> | nor A8 |
| 19:30 | <zcorpan> | A11 is wrong |
| 19:31 | <zcorpan> | A12 has bogus rationale |
| 19:32 | <zcorpan> | A13 seems backwards |
| 19:33 | <zcorpan> | A14 has bogus rationale |
| 19:57 | <Hixie> | BlueG: the mailman bug was fixed, you should be able to subscribe now |
| 22:20 | <zcorpan> | hsivonen: bookmarklet for v.nu: http://tinyurl.com/6jvpmv |
| 22:21 | <zcorpan> | works for both html and xhtml |
| 22:27 | <zcorpan> | http://tinyurl.com/65o6bz works with svg too |
| 22:30 | <zcorpan> | sorry, http://tinyurl.com/5mmt3l |
| 22:57 | <zcorpan> | should we expect episode 2 today? |
| 23:26 | <Hixie> | oh jesus |
| 23:26 | <Hixie> | kristof and garett talking to each other |
| 23:37 | <roc> | I thought that was part of your clever plan |
| 23:39 | <Hixie> | it does keep them occupied and away from discussing other things |
| 23:39 | <Hixie> | but i'm not sure it's wise |