| 00:03 | <zcorpan_> | http://validator.nu/?doc=http%3A%2F%2Fwww.mindfly.com%2F |
| 00:04 | <othermaciej> | hmm |
| 00:04 | <othermaciej> | all of his "Did you know..." questions seem false |
| 00:04 | <othermaciej> | in their premise |
| 00:04 | <othermaciej> | (except maybe Google Wave?) |
| 00:05 | <zcorpan_> | othermaciej: see Hixie's comment |
| 00:05 | <Hixie> | even his google wave one is wrong; see my comment |
| 00:06 | <Hixie> | (and btw the use of gears in wave will become a non-issue as soon as arun's draft is ready and browsers have drag-and-drop File objects) |
| 00:06 | <Midler> | isnt that the year w3c is having it to be a standard? |
| 00:06 | <Midler> | 2022? |
| 00:06 | <Hixie> | see my comment on the blog entry :-) |
| 00:07 | <ojan> | and FWIW, eric's patch to add drag-and-drop files to webkit came out of the Wave team's request so they would not have to rely on Gears |
| 00:10 | <Philip`> | jgraham: You should include some ASCII-art diagrams to break up the flow of your long emails |
| 00:11 | Hixie | looks through ben millard's interesting tables collection to get ideas for examples for the spec |
| 00:13 | <Hixie> | i need a table that has a scope=rowgroup header cell in a non-left-most cell |
| 00:15 | <othermaciej> | I feel like this guy doesn't understand how HTML works |
| 00:20 | <Midler> | iam still reading it |
| 00:21 | <Midler> | "The truth is that plugins can ‘upgrade the web‘ in under a year." |
| 00:21 | <Midler> | That doesnt sounds that fantastic |
| 00:21 | <zcorpan_> | Hixie: use google's index to find it |
| 00:21 | <Midler> | Like we will be using flash forever |
| 00:23 | <Hixie> | zcorpan_: hah |
| 00:23 | <Midler> | I must ask how much the possibility for huge changes in the html5 document is? |
| 00:24 | <Hixie> | Midler: if someone finds a huge problem, pretty high :-) |
| 00:24 | <Midler> | Ok |
| 00:24 | <Midler> | I read comments concerning that |
| 00:24 | <Midler> | On Swedish forums |
| 00:24 | <zcorpan_> | Midler: pointer? |
| 00:24 | <Midler> | Its like: Ok, mozilla firefox add html 5 support |
| 00:24 | <Philip`> | Hixie: I found a huge problem in the spec |
| 00:25 | <Philip`> | Hixie: It doesn't rhyme at all |
| 00:25 | <Midler> | but you would never know if that support will be left 5 years from now |
| 00:25 | <Hixie> | Philip`: good lord, you're right. i'll make it rhyme right away. |
| 00:27 | <zcorpan_> | Midler: features that are shipped in multiple browsers and used on a wide scale on the web are very unlikely to change in incompatible ways |
| 00:27 | <Midler> | "very unlikely" is the concern...you can never know |
| 00:29 | <Hixie> | who knows |
| 00:29 | <Hixie> | tomorrow maybe the entire internet infrastructure will collapse |
| 00:29 | <Hixie> | and there will be no web |
| 00:29 | <Hixie> | or maybe adobe will go bankrupt, and there will be no flash |
| 00:29 | <Hixie> | you can never know :-) |
| 00:29 | <Philip`> | I think Google should buy Adobe |
| 00:30 | <beowulf> | which group does Shelley Powers refer to when she describes them as 'not diverse'? |
| 00:30 | <zcorpan_> | Midler: do you have a link to said forums? |
| 00:31 | <zcorpan_> | Hixie: i guess there would still be flash if adobe went bankrupt |
| 00:31 | <Midler> | zcorpan_: http://www.idg.se/2.1085/1.236210/se-upp-flash--html-5-vill-vara-med-och-leka?articleRenderMode=listpostings#a19.651633 |
| 00:32 | <Midler> | Its not forum, bit comments |
| 00:32 | <othermaciej> | did Shelley re-un-quit? |
| 00:33 | <Dashiva> | Yes |
| 00:35 | <Midler> | zcorpan_: Swedish? |
| 00:38 | <Midler> | There is comment that says this: "the problem with W3C is that the whole standard in theory could be rewritten entirely, which would make that they do not have as good implementation of the longer." <-- slightly modified |
| 00:38 | <Midler> | a google translate* |
| 00:42 | <Dashiva> | Not quite |
| 00:43 | <Dashiva> | "Of course, it only applies to the candidate [spec of] CSS 2.1; the problem with w3c being so senselessly slow is that the entire standard could in theory be rewritten, which would make it so they don't have as good an implementation any more" |
| 00:50 | <Midler> | oh i see. Wrong of me |
| 00:51 | <Midler> | but reading the discussion show a concern about the time it takes for it to be a standard |
| 00:52 | <Midler> | "The web is evolving faster than w3c makes something a standard" |
| 00:54 | <othermaciej> | Midler: in theory it could be, but in practice it won't |
| 00:54 | <Midler> | =) |
| 00:54 | <Midler> | Even i understand that |
| 00:54 | <othermaciej> | Midler: once a spec is at the Candidate Recommendation phase, any significant changes require a lot of bureaucracy |
| 00:55 | <othermaciej> | the basic idea is not to lock it down until we have two interoperable implementations so we're sure it works |
| 00:55 | <othermaciej> | it's similar to the IET |
| 00:55 | <othermaciej> | IETF I mean |
| 00:55 | <othermaciej> | where in theory, a "Draft Standard" could be changed completely before becoming an "Internet Standard", but in practice it won't be |
| 00:56 | <jcranmer> | beowulf: ancient Saxony? |
| 01:23 | <Hixie> | jgraham: i crashed your table inspector :-( |
| 01:26 | <Hixie> | jgraham: http://junkyard.damowmow.com/391 |
| 01:27 | <Hixie> | jgraham: http://james.html5.org/tables/table_inspector.py?algorithm=html5c&input_type=type_uri&uri=http%3A%2F%2Fjunkyard.damowmow.com%2F391&source=%3Ctable%3E%3Cthead%3E%3Ctr%3E%3Cth%3E%3Cth%3E%3Ctbody%3E%3Ctr%3E%3Cth+scope%3Drowgroup%3E |
| 01:27 | <Hixie> | ->crash |
| 01:27 | <Hixie> | it works if i add the two missing cells that hsivonen's validator complains about |
| 01:37 | <zcorpan_> | Hixie: maybe you should have an example in the spec where a script moves the element that the parser is inserting to |
| 01:37 | <Hixie> | send mail :-) |
| 01:38 | <Hixie> | i'm busy trying to make exampels for tables |
| 01:54 | <zcorpan_> | hmm, i tried to annotate "Parsing HTML documents" but it added a box to #comments instead :-/ |
| 01:56 | <Hixie> | yeah the annotation thingy is broken |
| 01:56 | <Hixie> | not sure what the bug is |
| 01:56 | <Hixie> | haven't looked |
| 01:56 | <Hixie> | i think it has to do with the "impl" annotations |
| 02:28 | <heycam> | zcorpan_, fixed the [NoInterfaceObject] / [Constructor] thing, thanks |
| 02:33 | <zcorpan_> | heycam: nice |
| 02:35 | <zcorpan_> | Hixie: the parts of html5 that were split out to w3c docs don't have the same permissive license as html5 |
| 02:36 | <zcorpan_> | Hixie: so it's harder to fork web storage |
| 02:39 | tantek | prefers open standards specs that use a public domain declaration in order to be truly open. |
| 02:43 | <zcorpan_> | Hixie: you should have a third color for legs and tails |
| 02:44 | <zcorpan_> | Hixie: at first i thought the green arrows were all from Cats and English speakers |
| 03:23 | <othermaciej> | tantek: in practice, I don't think a public domain declaration is better than, say, a BSD-style copyright license or rough equivalent |
| 04:18 | <Hixie> | zcorpan: the source is still in the whatwg svn under that license :-) |
| 07:33 | <tantek> | othermaciej re: PD vs BSD etc. - in general I would agree, however, my experience with book publishers specifically for example is that they know/understand/like content that is cleared as PD but few have heard of BSD, and unknown = fear/uncertainty/doubt. |
| 07:50 | <Mrmil> | What is more correct? Open <code>bloodyscript.rb</code>, open <var>bloodyscript.rb</var> or open <bloodysomethingelse?>bloodyscript.rb</bloodysomethingelse?> |
| 07:55 | <jgraham> | So. Much. Mail. |
| 07:58 | <othermaciej> | Mrmil: I think <samp> would be most correct |
| 07:59 | <Mrmil> | othermaciej: Ook, thanks |
| 07:59 | <othermaciej> | no, I guess not, maybe <kbd> |
| 07:59 | <othermaciej> | HTML has too many elements to set things in monospace |
| 07:59 | <Mrmil> | right |
| 08:00 | <Mrmil> | <kbd>/var/www/test/script.rb</kbd>? |
| 08:00 | <Mrmil> | it's not too right though, it's more a path then input... but then you can't follow path without an input |
| 08:02 | <jgraham> | Mrmil: It is irelevent. No UA will ever care. People are really unlikely to screen scrape for it and if you care about people trying to you could always use <code class=path> or something |
| 08:02 | <Mrmil> | jgraham: ok, I generally use <code> anyways. I just wanted to satisfy my curiosity. :) |
| 08:13 | <sayrer> | Mrmil: if you want something to be monospace, you can use <font> |
| 08:15 | <Mrmil> | sayrer: I use css for that, I was curious if there is a way to correctly annotate files and paths, looks like there is not just one solution |
| 08:16 | <sayrer> | I see, I'm sure the annotation is important ;) |
| 08:18 | <Mrmil> | sayrer: :P |
| 08:30 | <hsivonen> | hmm. I've lost the URL of Hixie's <meta charset> tests again |
| 08:31 | <hsivonen> | http://hixie.ch/tests/adhoc/html/parsing/encoding/ |
| 09:18 | <MikeSmith> | Hixie: the hide/highlight feature doesn't work at all in webkit-based browsers, afaict |
| 09:18 | <Hixie> | wfm |
| 09:18 | <MikeSmith> | hmm |
| 09:18 | <Hixie> | what happens for you? |
| 09:18 | <Hixie> | i only tested it in webkit trunk |
| 09:18 | <Hixie> | but it doesn't do anything really fancy |
| 09:19 | <MikeSmith> | I'm testing in webkit nightly |
| 09:20 | <Hixie> | what does it do? |
| 09:20 | <Hixie> | what are you doing? |
| 09:20 | <MikeSmith> | the buttons work as expected, but the URLs don't seem to for me |
| 09:20 | <Hixie> | paste one here? |
| 09:20 | <MikeSmith> | http://www.whatwg.org/specs/web-apps/current-work/?style=author |
| 09:20 | <MikeSmith> | if I go to that URL, it still shows me the impl stuff |
| 09:21 | <Hixie> | hm, yeah |
| 09:21 | <Hixie> | the right radio button is checked though right? |
| 09:22 | <MikeSmith> | Hixie: yeah |
| 09:23 | <Hixie> | weird |
| 09:23 | <Hixie> | the styles are being disabled properly |
| 09:23 | gsnedders | wonders how long it'll take to buy tickets at King's Cross |
| 09:23 | <MikeSmith> | maybe webkit bug |
| 09:24 | <Hixie> | must be |
| 09:24 | <Hixie> | don't see what else it could be |
| 09:25 | <MikeSmith> | Hixie: want me to ask or on #webkit, or do you? |
| 09:26 | <Hixie> | go ahead |
| 09:27 | <MikeSmith> | k |
| 09:27 | <Hixie> | if the right radio is being checked, i don't see how it could fail |
| 09:31 | <zcorpan_> | hixie: multipage/styler.js is a 404 |
| 09:31 | <Hixie> | oops |
| 09:32 | <Hixie> | ok, updater.js and styler.js will be symlinked properly next time i regen |
| 09:36 | <Hixie> | heycam: did you come up with an attribute name for merging interfaces? |
| 09:49 | <heycam> | Hixie, haven't got to that one yet |
| 09:50 | <Hixie> | heycam: let's make one up for now, so i don't have to check in about six bazillion [XXX]s |
| 09:50 | <Hixie> | (i'm merging in the old DOM2 HTML deprecated stuff) |
| 09:51 | <heycam> | sure, just make something up and i'll let you know when i get to it |
| 09:51 | <Hixie> | ok |
| 09:53 | <Hixie> | i'll use [Supplemental] for now |
| 09:56 | <heycam> | righto |
| 10:33 | <gsnedders> | Philip`: ping |
| 10:36 | <gsnedders> | Philip`: You want to meet up at 17:00 or so today? |
| 10:36 | gsnedders | probably will be gone before you respond, but can probably lurk around at the normal place at that time anyway :P |
| 10:36 | <gsnedders> | (weee… multi-second lag!) |
| 10:37 | <hsivonen> | "In reality, version 3 of the iPhone OS is the only solid platform that has full HTML5 support (as well as some other fixed development platforms)." |
| 10:37 | <hsivonen> | http://www.davidtucker.net/2009/06/19/an-honest-open-discussion-on-web-standards-and-html-5/ |
| 10:38 | <Hixie> | i'm sure apple will be really glad to hear that they don't have to implement any more stuff |
| 10:38 | <Hixie> | still no acknowledgement of my comment on that post |
| 10:39 | gsnedders | really does have to go now, so maybe see you |
| 10:39 | <Hixie> | later |
| 10:39 | <gsnedders> | (Actually, I think you have my number, so phone me) |
| 10:40 | gsnedders | heads off to Cambridge |
| 10:58 | <zcorpan_> | Hixie: hmm, why is size DOMString on font bug long on basefont? |
| 10:58 | <Hixie> | dunno, but that's what DOM2 HTML says |
| 11:05 | <zcorpan_> | seems in ie both are long, in opera both are DOMString, while firefox and webkit follow dom2html |
| 11:27 | <hsivonen> | zcorpan_: hmm. Interesting bug about the errors for xmlns:dc being reported for href |
| 11:28 | <hsivonen> | btw, that's pretty certainly a useless use of RDFa |
| 11:28 | <hsivonen> | pure metacrap |
| 11:31 | <zcorpan_> | hsivonen: it's useful because it found a bug in your validator :) |
| 11:33 | <hsivonen> | :-) |
| 11:40 | <zcorpan_> | Hixie: the note about dom feature strings seems to be incorrect now |
| 11:41 | <zcorpan_> | Hixie: maybe you should just use "HTML", "2.0" |
| 11:42 | <Hixie> | send |
| 11:42 | <Hixie> | |
| 11:42 | <Hixie> | :-P |
| 11:42 | <zcorpan_> | working on it |
| 11:42 | <Hixie> | :-) |
| 11:44 | <hsivonen> | whoa. lots of email on public-html. |
| 11:44 | <zcorpan_> | hsivonen: probably partially my fault :( |
| 11:44 | hsivonen | has been happily looking away from public-html (towards the Mozilla land instead) today |
| 12:15 | <Lachy> | yeah, there's way too much e-mail on public-html |
| 12:16 | <Lachy> | I've got through about a third of this week's discussion, still have just over 100 to go |
| 12:32 | <Dashiva> | Last week was almost empty, though. Strange how it fluctuates. |
| 12:49 | <Lachy> | http://www.contemporis.com/2009/06/translation-from-adobe-pr-speak-to-english/ |
| 12:50 | <zcorpan_> | Lachy: that's copied from somewhere else |
| 12:51 | <zcorpan_> | http://news.ycombinator.com/item?id=662620 |
| 12:52 | <Hixie> | oops, marked my massive DOM2 HTML checkin as editorial! |
| 12:52 | <Hixie> | oh well |
| 12:53 | <Hixie> | i think that is it as far as the obsolete section goes, so if anyone knows of anything that should be in there, please file a bug or send mail |
| 12:53 | <Lachy> | zcorpan_, oh well. Wherever it originally came from, it's still somewhat funny. |
| 12:54 | <Lachy> | although, I'm curious about the claim "iPhone helped to radically increase the number of phones with Flash support" since the iPhone doesn't support Flash |
| 12:54 | <Hixie> | ok bed time |
| 12:54 | <Hixie> | nn |
| 13:01 | <MikeSmith> | zcorpan_, hsivonen - you guys around? |
| 13:02 | <zcorpan_> | MikeSmith: yes |
| 13:13 | <jgraham> | The local rate of public html emails is about 1 per 10 minutes. So making the reasonable assumption of a constant rate we should have about 600,000 emails between now and last call |
| 13:13 | <MikeSmith> | zcorpan_: sorry, got distracted |
| 13:13 | <MikeSmith> | zcorpan_: about http://bugzilla.validator.nu/show_bug.cgi?id=586 from hsivonen |
| 13:14 | <MikeSmith> | zcorpan_: is that not the same as http://bugzilla.validator.nu/show_bug.cgi?id=584 ? |
| 13:15 | <zcorpan_> | MikeSmith: seems so |
| 13:16 | <MikeSmith> | OK |
| 13:16 | <MikeSmith> | I thought at first that Henri was filing a new bug caused by my checkin |
| 13:16 | <MikeSmith> | zcorpan_: as far as I can tell, my checkin to the treebuilder code fixes that problem, without any regression |
| 13:17 | <zcorpan_> | MikeSmith: cool |
| 13:18 | <MikeSmith> | and checked in a change for the encoding-must-be-first-child thing yesterday too, so that should be fixed too, next time hsivonen deploys |
| 13:18 | <MikeSmith> | zcorpan_: it was just those two problems you found yesterday, right? |
| 13:20 | <zcorpan_> | MikeSmith: yeah |
| 13:20 | <MikeSmith> | great |
| 13:20 | <zcorpan_> | MikeSmith: in v.nu anyway |
| 13:21 | <MikeSmith> | zcorpan_: where else would it need to be fixed? |
| 13:21 | <zcorpan_> | i found and reported problems in the blog post itself, in wordpress, in mozilla, and in mozilla documentation |
| 13:23 | <zcorpan_> | actually, not mozilla |
| 13:26 | <zcorpan_> | sadly i didn't find problems in the spec |
| 13:37 | <MikeSmith> | zcorpan_: I see |
| 13:38 | <MikeSmith> | have we ever had any success in getting wordpress to make fixes? |
| 13:38 | <zcorpan_> | i noticed wordpress was being patched for <video> support already |
| 13:39 | <zcorpan_> | or at least to let it through their filter of whitelisted elements and attributes |
| 13:41 | <zcorpan_> | http://core.trac.wordpress.org/ticket/10151 |
| 13:42 | <zcorpan_> | http://core.trac.wordpress.org/ticket/9437 - case-insensitive svg in html ++ |
| 13:42 | <Dashiva> | <video src="spam-ad.ogg" autoplay /> |
| 13:42 | <Philip`> | zcorpan_: Hopefully not the whitelist for comments? |
| 13:43 | <Philip`> | Dashiva: That would be bad, because it would give browsers a disincentive to implement Ogg support |
| 13:43 | <zcorpan_> | Philip`: there were concerns about that in the bug |
| 13:43 | <Dashiva> | Sorry, spam-ad.mp4 of course |
| 14:51 | <Philip`> | Licensing is hard :-( |
| 14:53 | <Philip`> | I have some project that includes C++ source files, which will be GPL; and some art data (bitmap textures etc) which will be CC-by-sa; but then there's stuff like XML files that tie together the art data, so they're sort of art data but since they're XML they're sort of like code too, and it gets a bit confusing |
| 14:54 | <jgraham> | Philip`: Lets go shopping |
| 14:54 | <jgraham> | Or, if you prefer I will ask you a more interesting question |
| 14:54 | <jgraham> | Actually I will anywway |
| 14:55 | <jgraham> | (you can always stick your fingers in your eyes and hum) |
| 14:55 | <Philip`> | (That might hurt my eyes) |
| 14:55 | <jgraham> | Is there a good/standard way of empirically determining the big-O of a function? |
| 14:56 | <Philip`> | What do you count as "empirical"? |
| 14:56 | <jgraham> | At least under some reasonable set of assumptions e.g. the function has to be drawn from a finite set |
| 14:56 | <jgraham> | s/function/big O function/ |
| 14:57 | <zcorpan_> | http://forums.whatwg.org/viewtopic.php?t=4101 - spam or not? |
| 14:57 | <jgraham> | Philip`: Say I have an implementation of a "sort" function and I want to know if it is quadratic or subquadratic purely by black box measurement |
| 14:58 | <jgraham> | In a way that is reasonably robust against the environment |
| 14:58 | <Lachy> | JohnResig, I just sent a revised copy of the selectors api exit criteria to public-webapps, which also describes what should be done with the testsuite. Let me know what you think, and if and when you'll be able to make those changes |
| 14:59 | <jgraham> | zcorpan_: spam |
| 14:59 | <Philip`> | jgraham: It's not possible to determine a theoretically correct answer - the function could have cost n^1.5 + n^2/1000000000, and it would be O(n^2) but all your black box measurements would make you think it was O(n^1.5) |
| 15:00 | <Philip`> | (The definition of big-O notation is that it's all about what happens in the limit, and you can't perform measurements in the limit) |
| 15:01 | <jgraham> | Philip`: Indeed. So lets say that we don't care so much about the true asymptotic behaviour but the behaviour for some reasonable range of input values |
| 15:01 | <jgraham> | Which isn't the theoretically pure case but is more practically interesting |
| 15:02 | <Philip`> | Then you can just run it for a range of n values, and pass it into some statistical analysis package (e.g. Excel's graph plotter) and get it to tell you the polynomial equation of the curve |
| 15:02 | <jgraham> | Philip`: Right but does that seem likely to be robust in the face of environmental factors |
| 15:03 | <jgraham> | like, say, GC pauses or whatever |
| 15:03 | <jgraham> | and there are practical concerns about the range of input values that should be chosen |
| 15:03 | <Philip`> | (There's also the issue that big-O is technically about worst case behaviour, and e.g. quicksort has terrible worst-case behaviour but you'll never notice that except on carefully-chosen inputs) |
| 15:04 | <jgraham> | (Yeah the worst-case behaviour thing is a reasonable concern but we'll ignore that for now) |
| 15:04 | <Philip`> | jgraham: It seems likely to be sufficiently robust that the difference between n log n and n^2 would be blatantly obvious |
| 15:05 | <jgraham> | Philip`: OK but you are not aware of any useful prior art on doing this kind of meaurement? |
| 15:05 | <Philip`> | I don't see why it's a complicated enough issue that one would need prior art, rather than just measuring some values and looking to see what the curve is like :-) |
| 15:06 | <Philip`> | or looking at the implementation and working out what it's doing |
| 15:06 | <jgraham> | Philip`: I have been told that this is harder in practice than it sounds in theory :) |
| 15:07 | <Philip`> | (In the latter case, see e.g. Knuth for more information than you ever wanted to know about sorting algorithms) |
| 15:07 | <Philip`> | jgraham: I guess it depends on how precise an answer you want |
| 15:07 | <Philip`> | e.g. n log n vs n^2 is easy to detect, but n^1.99 vs n^2 would be much harder |
| 15:09 | <Philip`> | I suppose the problem is I don't understand what problem you're trying to solve |
| 15:11 | <jgraham> | Philip`: Fair enough :) |
| 15:12 | jgraham | isn't quite sure either |
| 15:12 | <Philip`> | Easy solution: Just use heap sort |
| 15:12 | <Philip`> | Uh, merge sort |
| 15:13 | <Philip`> | I think that's the one I mean |
| 15:15 | <zcorpan_> | smylers++ on UAs generating a description of the table structure |
| 15:16 | <jgraham> | Philip`: It isn't really a sort-specific question |
| 15:17 | <Philip`> | jgraham: Oh |
| 15:24 | <Philip`> | jgraham: Anyway, I think in conclusion I don't have any useful non-obvious knowledge that would help you do whatever it is you're trying to do |
| 15:27 | <JohnResig> | Lachy: k - I'll probably be able to look at it/make the changes early July |
| 15:35 | <Lachy> | ok |
| 15:37 | <zcorpan_> | http://blog.benward.me/post/128294973 |
| 15:41 | <jgraham> | I think it's sweet of Adobe to talk about HTML5 at every possible oppertunity. |
| 15:41 | <jgraham> | It's hard to imagine what more they could do to promote it |
| 15:44 | <Philip`> | They could make Flash buggy and crash-prone, and not port it to new platforms and devices |
| 15:45 | <jgraham> | Philip`: Like it said it's hard to imagine what more they could do |
| 15:45 | <jgraham> | s/it/I/ |
| 15:59 | <zcorpan_> | jgraham: do you think it would be possible to hack together a demo implementation of generating a description of a table's structure? |
| 16:00 | <jgraham> | zcorpan_: All things are possible |
| 16:00 | <zcorpan_> | jgraham: i mean, do you want to do it? |
| 16:00 | <jgraham> | zcorpan_: I can have a go |
| 16:00 | <zcorpan_> | cool |
| 16:02 | <jgraham> | zcorpan_: Another interesting project would be to see how well the Mozilla layout table detector works |
| 16:02 | <zcorpan_> | yeah |
| 16:02 | <zcorpan_> | i wonder how mozilla's algorithm came into being |
| 16:02 | <Lachy> | are you show that it's possible for UAs to determine for itself what would otherwise be provided by a summary attribute (if one is supplied and used properly)? |
| 16:03 | <jgraham> | If it works reasonably well hsivonen could add it to the validator |
| 16:03 | <jgraham> | Lachy: No. But it seems worth finding out |
| 16:03 | <zcorpan_> | is it just a bunch of guesses? or is it based on research? |
| 16:03 | jgraham | guesses it is a bunch of guesses |
| 16:03 | <Lachy> | jgraham, certainly, if we could show that summary was redundant, then it would end the debate |
| 16:04 | <Philip`> | Lachy: I assume that was a joke |
| 16:04 | <jgraham> | Lachy: Hah ha |
| 16:05 | <Lachy> | no, it wasn't a joke. I'm serious. If everything an author is supposed to provide with a summary attribute can be determined automatically from the table structure, then it is redundant |
| 16:05 | <zcorpan_> | jgraham: look at ben's collection of tables for inspiration |
| 16:06 | <jgraham> | Lachy: I doubt it will be possible to determine "everything an author is supposed to provide" |
| 16:06 | <jgraham> | Even if it is I think people will still claim @summary is needed |
| 16:07 | <jgraham> | (for legacy UAs or for super-complex corner cases or something) |
| 16:07 | <Philip`> | Is summary="Calendar" considered a useful use of summary? That's something that's very hard to determine algorithmically, and hard for someone using a non-graphical browser to determine, but trivial for someone using a graphical browser |
| 16:08 | <jgraham> | Philip`: That's not what @summary is for, apparently |
| 16:08 | <Lachy> | Philip`, I don't think so |
| 16:08 | <jgraham> | (my current understanding is that it is _only_ supposed to be used for describing the structure of the table) |
| 16:08 | <Philip`> | jgraham: It doesn't really matter what it's apparently for, it matters whether it's beneficial to users |
| 16:09 | <zcorpan_> | would be interesting to see if it's possible to determine whether a given table is a calendar or not |
| 16:09 | <zcorpan_> | calendars usually have a specific format |
| 16:09 | <Lachy> | "This attribute provides a summary of the table's purpose and structure ..." - For most common simple cases, it's possible to determine the structure simply by looking at the headings. The purpose should be indicated by the caption or surrounding prose. |
| 16:15 | <beowulf> | if the ua can't provide a summary, allow the author to use @summary or whatever, no? |
| 16:16 | <Lachy> | beowulf, even if the UA can't, there are better proposals than the @summary |
| 16:42 | <remysharp> | hi - I've been playing with the manifest for offline apps in Safari, and checking the docs because Safari won't reload a server side change - is this being handled correctly? |
| 16:42 | <remysharp> | i.e. the page is loaded and told to cache xyz assets |
| 16:42 | <remysharp> | I make a change to the page, like the h1 text |
| 16:42 | <remysharp> | and Safari doesn't reflect the change |
| 16:42 | <remysharp> | I can only get it to load the new page if I cache bust the URL manually. |
| 16:43 | <ap> | remysharp: you need to change the manifest for resources to be refreshed |
| 16:43 | <remysharp> | but it was the actual page that was being loaded, i.e. it's not specified in the manifest |
| 16:44 | <ap> | remysharp: it goes to the manifest implicitly (as a master resource) |
| 16:44 | <remysharp> | aye, that's what I figured - |
| 16:44 | <remysharp> | are you saying if I 'touch' the manifest, it should update it? |
| 16:45 | <remysharp> | or - something else, how should I be able to tell the browser the master resource has been updated? |
| 16:45 | <ap> | remysharp: its content should change, not just modification date |
| 16:45 | <ap> | remysharp: its == manifest |
| 16:45 | <remysharp> | but I've got nothing to add though - it's just a change to the html |
| 16:45 | <ap> | remysharp: you can add a version in a comment |
| 16:45 | <remysharp> | ah - right. |
| 16:45 | <remysharp> | so comments can trigger an update then? |
| 16:46 | <ap> | remysharp: any change to the manifest content will |
| 16:46 | <remysharp> | hm - just tried it - it didn't. |
| 16:46 | <remysharp> | but I guess that's an implementation bug, right? |
| 16:46 | <ap> | remysharp: why do you think so? |
| 16:47 | <remysharp> | because I changed the manifest, as you suggested |
| 16:47 | <remysharp> | and the rendered html didn't update in safari |
| 16:47 | <ap> | remysharp: so far, so good |
| 16:47 | <ap> | remysharp: changing the manifest and reloading should start an update process |
| 16:47 | <remysharp> | yeah, but it didn't. |
| 16:47 | <remysharp> | at all. |
| 16:48 | <ap> | remysharp: once it's done, and the cache is updated, your page gest an event |
| 16:48 | <remysharp> | it's not though |
| 16:48 | <ap> | remysharp: is is_not_ magically reloaded again |
| 16:48 | <remysharp> | I'm listening for all the events |
| 16:48 | <remysharp> | and I've also triggered manual update() |
| 16:48 | <ap> | remysharp: what event listeners do you have installed? |
| 16:48 | <remysharp> | and swap |
| 16:48 | <remysharp> | one sec |
| 16:48 | <remysharp> | checking,error,noupdate,downloading,progress,updateready,cached,obsolete |
| 16:48 | <remysharp> | that lot! |
| 16:48 | <ap> | remysharp: which ones did fire? |
| 16:49 | <remysharp> | currently - after manually refreshing - it's on 'downloading' |
| 16:49 | <remysharp> | but it's stuck on downloading, and the master resource just isn't updating so far. |
| 16:50 | <ap> | remysharp: sounds strange. it is possible that there's a bug, but we do have tests verifying the behavior |
| 16:50 | <remysharp> | hmm - and now Safari crashed! |
| 16:50 | <remysharp> | are these the ones that drive the test data on the whatwg specs? |
| 16:50 | <ap> | remysharp: all right, that's a bug :) |
| 16:50 | <remysharp> | because I've been writing a bunch of simple tests myself, and a few of them definitely don't marry up. |
| 16:51 | <zcorpan_> | hsivonen, MikeSmith: maybe all obsolete features in html should be in the html5 schema (including framesets), with assertions saying that it's obsolete and what to use instead |
| 16:51 | <remysharp> | for instance, drag and drop, from what I can see in testing - is only supported in Firefox - and not Webkit |
| 16:51 | <ap> | remysharp: WebKit tests are just for WebKit regression testing (we're getting offtopic here, could move to #webkit) |
| 16:51 | <remysharp> | ap: aye, sorry. |
| 16:52 | <remysharp> | ap: so, to trigger an update to the application cache, from the server, I should be updating the manifest - that's the approach |
| 16:52 | <ap> | right |
| 16:52 | <remysharp> | ap: is that in the spec - I didn't come across it when I was trying to see if I had to manually trigger it somehow. |
| 16:52 | <remysharp> | (I mean, I assume it is, and I missed it) |
| 16:54 | <ap> | remysharp: see step 7 in 6.9.4 |
| 16:54 | <MikeSmith> | zcorpan_: I have a patch to add frameset written up already |
| 16:54 | <zcorpan_> | MikeSmith: sweet |
| 16:55 | <ap> | remysharp: update process is invoked in many circumstances (e.g. each time a new page opens), and usually quickly finishes due to the manifest being unchanged |
| 16:55 | <MikeSmith> | frameset and basefont are the only obsolete things that aren't already in teh v.nu schemas |
| 16:55 | <ap> | remysharp: please file bugs at bugs.webkit.org about the crash, and about other issues you may find |
| 16:55 | <remysharp> | ap: yeah, the new page opens - that's what I was figuring on - but it doesn't seem to happen in webkit - re: bugs, will do - if I can consistently create them. |
| 16:56 | <remysharp> | ap: re: 6.9.4 is that on www, or dev.w3 ? I'm on www and offline is in 5.x |
| 16:56 | <zcorpan_> | MikeSmith: is marquee in? |
| 16:56 | <ap> | remysharp: I was looking at whatwg.org |
| 16:56 | <remysharp> | ap: ah! |
| 16:56 | <zcorpan_> | MikeSmith: spacer? |
| 16:57 | <MikeSmith> | zcorpan_: nope. I meant elements that were valid in HTML4 but that are not valid in the HTML5 draft |
| 16:57 | <zcorpan_> | MikeSmith: ok. i'm looking at the obsolete section in html5 |
| 17:06 | <zcorpan_> | MikeSmith: is it possible to generate messages that say what to use instead for obsolete attributes? |
| 17:12 | <MikeSmith> | zcorpan_: yeah, it's possible to do with the assertions-checking code |
| 17:12 | <remysharp> | ap: cheers for the help. |
| 17:12 | <MikeSmith> | zcorpan_: anything is possible with that. we can make the error messages say whatever we want |
| 17:13 | <zcorpan_> | MikeSmith: i think it would be useful to rip off what Hixie wrote in the obsolete section |
| 17:14 | <zcorpan_> | MikeSmith: hmm, actually |
| 17:14 | <zcorpan_> | MikeSmith: some messages already have the content model stuff |
| 17:15 | <MikeSmith> | zcorpan_: yeah, for some cases |
| 17:15 | <zcorpan_> | MikeSmith: maybe it would make sense to include a separate field in a similar manner for advice on what to use instead |
| 17:15 | <zcorpan_> | MikeSmith: instead of adding everything to the schema and assertions |
| 17:17 | <MikeSmith> | we can't add arbitrary messages to the grammar-based schema checking anyway. we are stuck with what Jing reports there, for the most part |
| 17:17 | <MikeSmith> | the content-model stuff that hsivonen has it emitting now is something that he hooked in with custom code |
| 17:18 | <MikeSmith> | it just copies that from a copy of the spec, verbatim |
| 17:18 | <zcorpan_> | MikeSmith: can't the advice be implemented the same as the content model stuff? |
| 17:19 | <MikeSmith> | it could be, but it would be more trouble than just adding it to the assertions checking |
| 17:19 | <zcorpan_> | ok |
| 17:19 | <MikeSmith> | the assertions checking is now implemented as custom Java code anyway |
| 17:19 | <MikeSmith> | so we can do anything we want with that |
| 17:19 | <MikeSmith> | there is still the assertions.sch file, but v.nu doesn't actually use that |
| 17:20 | <zcorpan_> | aha |
| 17:20 | <MikeSmith> | anyway, v.nu has an mechanism already for emitting Info-class messages (as opposed to Error or Warning) |
| 17:20 | <MikeSmith> | so that part's already in there |
| 17:21 | <zcorpan_> | i was having in mind including the advice in the same error message |
| 17:21 | <zcorpan_> | like content model stuff is tied to a an error message |
| 17:22 | <MikeSmith> | I think it would need to be Info so that users can suppress it if they want |
| 17:22 | <MikeSmith> | IMHO, that's how the content-model stuff should be too |
| 17:22 | <MikeSmith> | they can still be tied together |
| 17:23 | <MikeSmith> | could just think of them as different fields of the same message |
| 17:24 | <MikeSmith> | anyway, I chatted with hsivonen a while back of using the assertions-checking backend for more cases, particularly cases where the error messages from Jing aren't very useful |
| 17:24 | <MikeSmith> | which is actually a lot of cases |
| 17:24 | <zcorpan_> | MikeSmith: some parse errors already have advice on what to do instead |
| 17:25 | <MikeSmith> | yeah, the error messages for parse errors is another separate thing |
| 17:27 | <MikeSmith> | as far as the schema-checking part goes, I think one lesson from this is that you can't ever expect to get useful messages for all cases from a grammar-based generic checker like Jing or any other RelaxNG or XSD checker |
| 17:28 | <MikeSmith> | if you want good error messages for more cases, it has to be supplemented with other kinds of checking |
| 17:28 | <MikeSmith> | schematron assertion-based checking is actually a really good way |
| 17:29 | <MikeSmith> | but the performance sucks |
| 17:30 | <MikeSmith> | and it is based on XPath, and not everybody likes XPath, and writing XPath expressions for some cases is a PITA |
| 17:31 | jgraham | onders how the fraction of custom java code in validator.nu has changed as a function of time |
| 17:31 | <jgraham> | +w |
| 17:31 | <zcorpan_> | so the conclusion is to use relax ng for anything that doesn't need useful messages and custom code for the rest? |
| 17:32 | <MikeSmith> | jgraham: the custom code for assertion checking has been going down recently because Hixie has relaxed or removed some constraints |
| 17:32 | <MikeSmith> | like the one about encoding declaration needing to be first child of head |
| 17:32 | <MikeSmith> | or around datagrid content model |
| 17:33 | <MikeSmith> | the datagrid content-model checking was a big chunk of the assertions code |
| 17:33 | <MikeSmith> | zcorpan_: no, not exactly -- not IMHO at least |
| 17:33 | <MikeSmith> | though that is basically what v.nu is doing now |
| 17:34 | <MikeSmith> | well, not exactly |
| 17:34 | <MikeSmith> | lemme rephrase that |
| 17:35 | <MikeSmith> | v.nu now is mostly using relax ng for anything that can be expressed practically in relax ng, and then using the assertions checking for stuff that can't, as then other custom code for stuff that can't be done with assertions checking (e.g., overlapping table cells) |
| 17:36 | <MikeSmith> | so relax ng does emit useful messages for a lot of cases |
| 17:36 | <MikeSmith> | but there are some others where it doesn't -- like cases that Rick Jelliffe mentioned |
| 17:37 | <MikeSmith> | e.g., where the constraint is around allowed sequences of elements |
| 17:47 | webben_ | notes that one major advantage of DTD-based validation is developers can hack the DTD to change the constraints. I don't think that offsets the way DTDs limit what you can check, but it is a major advantage nevertheless. |
| 17:48 | <webben_> | e.g. at Yahoo!, we're on route to using a custom DTD internally for validating the sub- and super-sets of HTML4 we want to be using. |
| 17:48 | <brale> | hi all. I need some help, an idea. I need to extract only (printable) text from some page. Is there an easy way to do this using html5lib or some other library? |
| 17:50 | <zcorpan> | webben_: you can hack a relax ng schema, too |
| 17:51 | <webben_> | zcorpan: Oh absolutely, and I much prefer that. |
| 17:52 | <webben_> | The contrast is more with arbitrary code vs declarative rulesets. |
| 17:52 | <webben_> | so we're on route to a custom DTD, but the amount of effort it would require to customize the CSS validator is much higher |
| 17:53 | <webben_> | so there our immediate approach is to try to comment out bits of CSS we know won't validate |
| 17:53 | <webben_> | which is problematic in itself. |
| 17:58 | <Philip`> | webben_: Might it be easier to filter the output of the CSS validator, to remove the errors you don't care about? |
| 17:59 | <webben_> | Philip`: There are some problems with that. |
| 17:59 | <webben_> | some errors will cause the validator to stop parsing, or (iirc) parse differently than the parsing specified in CSS2.1 |
| 17:59 | <Philip`> | Oh, that doesn't sound good |
| 18:00 | <webben_> | other errors that would work for - e.g. I think that would work for unrecognized properties. |
| 18:00 | <atwilson> | Hixie: Is it our intent to eventually expose WebSockets to workers? |
| 18:04 | <webben_> | Philip`: contrast the output for a{*height:50px;min-height:50px;}a{color:red;} and a{height:50px;min-height:50px;}a{color:red;} |
| 18:05 | <webben_> | Philip`: in the first case, W3C validator implies there's no "Valid CSS information" |
| 18:05 | <webben_> | it recovers from a{foo:50px;min-height:50px;}a{color:red;} rather differently |
| 18:11 | <jgraham> | brale: Yes that is possible |
| 18:11 | <jgraham> | How easy it is depends on exactly what you want |
| 18:13 | <jgraham> | e.g. import html5lib;import urllib2;from lxml import etree;print etree.tostring(html5lib.parse(urllib2.urlopen(url)), method="text") |
| 18:13 | <jgraham> | er |
| 18:13 | <jgraham> | import html5lib;import urllib2;from lxml import etree;print etree.tostring(html5lib.parse(urllib2.urlopen(url), treebuilder="lxml"), method="text") |
| 18:14 | jgraham | guesses he made a dozen other mistakes too |
| 18:20 | <brale> | jgraham: thanks. that was something I was thinking about (only i would use text_content). The thing is that I need only text that would actually be shown on the screen |
| 18:21 | <brale> | jgraham: so I am hoping that there is a more elegant solution to my problem. The main problem with lxml is that i will get text that's inside <script></script>, etc. |
| 18:23 | <jgraham> | brale: Yeah you would need to blacklist some elements. If you also care about the style rules that are in effect hings become much more complex |
| 18:24 | <zcorpan> | brale: hmm, <script> is something that http://lists.w3.org/Archives/Public/public-html/2009Jun/0682.html maybe should deal with too |
| 18:24 | <brale> | zcorpan: i think it would be smart |
| 18:25 | <zcorpan> | brale: what's your use case? |
| 18:25 | <brale> | jgraham: I don't care about style, I'm only interested in text |
| 18:26 | <brale> | zcorpan: text-mining, statistical models |
| 18:27 | <zcorpan> | brale: other than <img>, <script> and <style>, do you know which elements should be special-cased? |
| 18:29 | <brale> | zcorpan: <video> and some other html5 tags |
| 18:29 | <zcorpan> | how would you special case <video>? |
| 18:30 | <brale> | zcorpan: well I dont need anything thats inside |
| 18:30 | <brale> | as it would not give my any text on the screen |
| 18:32 | <zcorpan> | brale: would you want the text in <img alt="...">? |
| 18:32 | <brale> | zcorpan: I suppose not |
| 18:34 | <brale> | maybe I should stop searching and start coding with html5lib/lxml and deal with problems as they come :D |
| 18:41 | <zcorpan> | smylers++ |
| 18:42 | <brale> | zcorpan, jgraham: thanks for help |
| 18:48 | <jgraham> | smylers keeps posting good stuff today. I hope it has more success than my attempt to explain why there might be disagreement |
| 18:50 | <jgraham> | Which seemed to fail. I'm not sure if it's because a) it was actually not that insightful b) it was not as well reasoned as I had hoped c) everything I say is regarded as tainted or d) other |
| 18:53 | <takkaria> | I suspect it's one of them |
| 18:55 | <Philip`> | jgraham: You could empirically test option c by adopting a fake name when posting to the list, so nobody will realise it's really you |
| 19:13 | <brale> | jgraham: do you now anything about lxml/html5lib error "Invalid attribute name u'xml:lang'" ? |
| 19:17 | <zcorpan> | brale: is it an lxml error or html5lib error? |
| 19:24 | <brale> | zcorpan, jgraham: Now I'me using html5lib from mercurial and it seems to work :D |
| 19:38 | <brale> | jgraham, zcorpan: as I see there is a small bug in current mercurial version in html5lib/treewalkers/_base.py |
| 19:38 | <brale> | jgraham, zcorpan: in function endTag, missing "," |
| 19:39 | <jgraham> | brale: The treewalkers stuff on head is probably slightly broken |
| 19:39 | <zcorpan> | jgraham: is the foreign branch still more up to date? |
| 19:40 | <jgraham> | zcorpan: No they have been merged |
| 19:40 | <zcorpan> | ok |
| 19:40 | <jgraham> | brale: I will check if I have any local patches that I should push out later but I am going out for a while now |
| 19:42 | <brale> | jgraham: ok. ( missing "," isn't that big ;) ) |
| 19:42 | <jgraham> | brale: Patches welcome :) |
| 19:43 | <brale> | jgraham: will html5lib.HTMLParser be fine for using? |
| 20:03 | Dashiva | wonders if debate about accessibility is much different from debate about religion |
| 20:07 | <Philip`> | Dashiva: It is, because accessibility has some underlying quantitative facts (like how many people can read how many sites), and the debate might lead to a better understanding of those facts |
| 20:16 | <Dashiva> | Philip`: This is true. But religion also has some underlying facts (history, written works, etc) that could be argued objectively. The problem is the people who turn every discussion into rhetorics. |
| 20:19 | <tantek> | Dashiva, I think the bigger problem is people who say "this is the way it is because I say so and I'm a [self-proclaimed] expert" - in other words, people that ask you to accept their assertions on faith. |
| 22:07 | <Hixie> | atwilson: yes |
| 22:07 | <atwilson> | OK, thx. |
| 22:24 | <zcorpan> | marginheight="" seems to be a way for pages to do cross-document communication, at least in one direction |
| 22:25 | <zcorpan> | cross-origin even |
| 22:26 | <Hixie> | ? |
| 22:26 | <Hixie> | oh, yeah, i guess |
| 22:26 | <Hixie> | heh |
| 22:36 | <zcorpan> | woot, <pre width> is not in the rendering section |
| 22:38 | <zcorpan> | though it seems only opera and firefox support it |
| 22:47 | <Hixie> | for those of you who were worried about paying for tpac using paypal, btw, apparently they will also accept checks made payable to MIT |
| 23:29 | <zcorpan> | http://wiki.whatwg.org/wiki/Presentational_elements_and_attributes |
| 23:33 | zcorpan | realizes that he uses border="1" on that page, but it seemed that class="wikitable" did not result in anything |