| 01:08 | <Hixie> | Philip`: yt? |
| 01:10 | <Philip`> | Hixie: Yes |
| 01:10 | <Philip`> | (though somewhat tired and not able to think perfectly coherently) |
| 01:10 | <Hixie> | i'm going through canvas feedback |
| 01:10 | <Hixie> | and got to your comment about 0x0 patterns |
| 01:11 | <Hixie> | and i realised that the spec doesn't say how to handle broken images as patterns either |
| 01:11 | <Hixie> | so i'm just gonna make both do the same thing |
| 01:11 | <Hixie> | i'm debating raise an exception, and return a pattern that is indistinguishable from a 1x1 transparent black pixel |
| 01:11 | <Philip`> | Broken images should have .complete = false and so it'll throw an exception if you try using them |
| 01:12 | <Hixie> | oh oops, it is defined. i missed that second paragraph. duh. |
| 01:12 | <Hixie> | well then i'll just make 0x0 canvases do that |
| 01:12 | <Hixie> | but we have other feedback to the effect that we should make fewer things raise exceptions |
| 01:12 | <Philip`> | Sounds reasonable |
| 01:12 | <Hixie> | so what "that" is might change |
| 01:14 | <Hixie> | safari seems to treat createPattern() with a non-complete image as an infinite black pattern (same as 'black') |
| 01:14 | <Philip`> | Is it not just ignoring the attempt to set fillStyle? |
| 01:15 | <Philip`> | I though it returned undefined from createPattern, or something like that |
| 01:15 | <Hixie> | ah yes |
| 01:15 | <Hixie> | indeed |
| 01:15 | <Philip`> | s//t/ |
| 01:15 | <Hixie> | it does that with a 0x0 canvas too |
| 01:15 | <Hixie> | k |
| 01:21 | <Hixie> | well the mozilla guys are against not raising exceptions for exceptional errorneous inputs |
| 01:21 | <Hixie> | so |
| 01:22 | <Philip`> | I thought the specific complaints about exceptions were things like negative-sized rectangles, where the spec currently throws exceptions but it's sensible to define non-erroneous behaviour |
| 01:22 | <Hixie> | yeah |
| 01:22 | <Hixie> | that will certainly be changed |
| 01:22 | <Philip`> | (as opposed to cases which are clearly errors and can't do anything sensible) |
| 01:22 | <Hixie> | right |
| 07:55 | <MikeSmith> | wondering if anybody's see this: |
| 07:55 | <MikeSmith> | http://timepedia.blogspot.com/2008/01/chronoscope-demo-in-flash-whatwg-canvas.html |
| 09:13 | <krijnh> | annevk: zet je presentatie van 8 feb ook online? |
| 12:43 | <zcorpan_> | 0004 / 345 Bad value “en-GB-hixie” for attribute “lang” from namespace “http://www.w3.org/XML/1998/namespace” on element “html” from namespace “http://www.w3.org/1999/xhtml”: Bad language tag: Bad variant subtag. |
| 12:48 | <zcorpan_> | Bad value “image” for attribute “type” on element “input” from namespace “http://www.w3.org/1999/xhtml”. seems like a confusing message |
| 12:49 | <Dashiva> | annevk, zcorpan_, Lachy: Add me to your default acid3 cc list, would you? :) |
| 12:49 | <hsivonen> | zcorpan_: or that a bug in v.nu? |
| 12:50 | <hsivonen> | is type='image' still in? |
| 12:50 | <zcorpan_> | it complains about either value='' or type='' when both are set, depending on which came first |
| 12:51 | <hsivonen> | fail-fast behavior :-) |
| 12:55 | <hsivonen> | hmm. perhaps I need to do some trick that reorders attributes before the RELAX NG stage |
| 12:56 | <zcorpan_> | i can't find where it says that image inputs can't have value |
| 12:57 | <Lachy> | Dashiva, what acid3 cc list? |
| 12:57 | <Lachy> | do you mean when we mail Hixie about bugs in the tests or something? |
| 12:59 | <zcorpan_> | ah, it doesn't apply according to the attribute summary in wf2, so i guess the prose bans it somewhere |
| 13:15 | <hsivonen> | hmm. someone has typoed and used a ns URI: http://www.w3.org/XML/1998/namespaces |
| 13:15 | <hsivonen> | that one would be hard to spot if V.nu allowed unknown namespaces to pass silently |
| 13:16 | <zcorpan_> | that might have been me :) |
| 13:16 | <zcorpan_> | if it was via the textarea interface |
| 13:17 | <hsivonen> | zcorpan_: it probably was. I don't log errors from POSTs, but it looks like I've accidentally logged unknown NS URIs from POSTed stuff |
| 13:18 | <zcorpan_> | (which wasn't a typo but experimenting) |
| 13:20 | <Philip`> | http://www.ninebynine.org/Software/HaskellUtils/HXmlToolbox-4.00/hdom/Namespace.hs - other people have made that typo |
| 13:21 | <zcorpan_> | hsivonen: it would be hard to spot only if the prefix was typoed, too |
| 13:22 | <zcorpan_> | since xmlns:xml="foo" is namespace-malformed |
| 13:25 | <zcorpan_> | hmm http://www.google.com/search?q=www.w3.org-1998-XML-namespace |
| 13:28 | <hsivonen> | POSTed unknown namespaces are no longer logged. |
| 13:51 | Philip` | sees that http://gnuplot.sourceforge.net/demo_svg/ uses <embed src=....svg>, and it doesn't work in Opera 9.2 (but does in 9.5 and FF2) |
| 13:57 | <hsivonen> | Philip`: with plug-in or without? |
| 13:58 | <Philip`> | hsivonen: No plugins |
| 13:58 | <hsivonen> | ok. |
| 13:59 | hsivonen | finds Verified Download Plug-in and Digital Rights Management Plugin on his plugi-in list |
| 14:04 | <Philip`> | Hmm, apparently I have plugins for QuickTime, RealPlayer, DivX and Windows Media |
| 14:04 | <Philip`> | seemingly due to mplayer pretending to be all of those |
| 18:18 | <Hixie> | if anyone wants to try to reply to this rant: http://blogs.techrepublic.com.com/programming-and-development/?p=599 |
| 18:18 | <Hixie> | ..then please be my guest |
| 18:18 | <Hixie> | i can't work out what to say, he doesn't make any specific criticisms |
| 18:24 | SadEagle | chuckles at the "impossible to print" part |
| 18:25 | <gavin> | he seems to think that the "issues that [proprietary] technologies try to solve" are "become difficult for the disabled, hard for search engines to parse, and impossible to print"? |
| 18:25 | <gavin> | that argument makes no sense |
| 18:29 | <Dashiva> | Well, impossible to print is probably DRM :) |
| 18:30 | <SadEagle> | gavin: it makes a bit more sense later. I guess it's an issue of what he views html as for --- as a pure document display/markup language. |
| 18:32 | <Dashiva> | But HTML fails at that too, compared to ps and pdf |
| 18:34 | <Dashiva> | He clarifies in a comment that he didn't want HTML4 changed at all (but he likes video and audio. |
| 18:35 | <SadEagle> | Well, I'd personally much prefer a less ambitious HTML5, but that doesn't make HTML5 -bad-. People have different goals. |
| 18:38 | <Dashiva> | And the web is moving away from static documents, pretending a spec can change that isn't very effective |
| 18:44 | <Hixie> | hsivonen: dude that is the most awesome data ever. well, for this week anyway. |
| 18:45 | <Hixie> | hsivonen: please do send more data on this at some point :-) |
| 19:18 | <hsivonen> | Hixie: thanks. I think I'll send some of my conclusions to the list as well. and yes, I intend to rerun the numbers later |
| 19:19 | <Hixie> | cool |
| 19:20 | <Hixie> | that e-mail alone has done more to convince me that we need to add Content-Type as a pragma to <meta> than any arguments up to this point |
| 19:20 | <Hixie> | not that i didn't agree with it before, but data is convincing. |
| 20:02 | jgraham_ | reads the techrepulic article and thinks for King Canute |
| 20:03 | <Dashiva> | I notice one new reply suggests doing away with scripting entirely |
| 20:05 | <mpt> | I'm sure the XHTML2 WG would welcome new contributors |
| 20:06 | <jgraham_> | "In other words, no XHTML 2, just an XML representation of HTML 5. Bleh." |
| 20:07 | <jgraham_> | He can't join the XHTML2 WG because he doesn't believe it exists :) |
| 20:09 | <hsivonen> | that page has more than one body start tag... |
| 20:33 | <jruderman_> | Hixie: i noticed that google search often returns sites with invalid certs. do you know whether the spider "knows" which sites have invalid certs, or just ignores certs entirely? |
| 20:34 | <jruderman_> | Hixie: i'm curious because some people at mozilla wanted data on how many public-facing sites have invalid certs |
| 20:34 | <Hixie> | no idea |
| 20:36 | <jruderman_> | "HTML 5 takes this smart direction, locks it in a warehouse full of gasoline and ball bearings, and throws a match inside." sounds like a perfectly specific criticism. why are you resorting to arson? |
| 20:36 | <Hixie> | have you any idea how hard it is to destroy tag soup? |
| 20:36 | <Hixie> | heavy fire is the only option |
| 20:37 | <Hixie> | not sure where he got the ball bearings from though |
| 20:38 | <Hixie> | why would we want to use ball bearings to burn something |
| 20:38 | <Hixie> | that seems weird |
| 20:38 | <jruderman_> | hmm, maybe that was his criticism. why *are* you using ball bearings when gasoline alone would do? |
| 20:39 | <Hixie> | we're not! |
| 20:39 | <Hixie> | doing so would be silly |
| 20:39 | <Dashiva> | Maybe the ball bearings are needed to keep the soup inside |
| 20:44 | <Hixie> | the tag soup is pretty viscous |
| 20:44 | <Hixie> | i don't think ball bearings would help |
| 20:46 | jwalden | thinks it's just the mainstream-ization of IEDs |
| 20:46 | <Hixie> | IEDs? |
| 20:47 | <gsnedders> | I thought the gasoline was just there to lubricate the ball bearings… |
| 20:47 | <jwalden> | improv explosive devices, popularized by an insurgency in a certain Middle-Eastern state? |
| 20:47 | <Hixie> | why throw a match then? |
| 20:47 | <gsnedders> | Hixie: I dunno, some other person firelighting for fun> |
| 20:47 | <gsnedders> | *? |
| 20:49 | <Hixie> | fair enough |
| 20:49 | <Hixie> | still a very confusing review |
| 20:59 | <zcorpan_> | speaking of confusing, can someone make sense of this? http://forums.whatwg.org/viewtopic.php?t=138 |
| 21:03 | kingryan | is getting an "internal server error" |
| 21:04 | <zcorpan_> | hmm, me too. worked when i posted the link |
| 21:04 | <zcorpan_> | now it works again |
| 21:06 | <hsivonen> | Hixie: btw, I reran the numbers. I haven't written my commentary for the numbers, yet, though |
| 21:07 | <hsivonen> | Hixie: the new numbers are in the same location |
| 21:07 | <hsivonen> | now with 400 distinct URIs |
| 21:07 | <Hixie> | uri? |
| 21:08 | <hsivonen> | Hixie: http://hsivonen.iki.fi/test/moz/analysis.txt |
| 21:09 | <zcorpan_> | only 2 had duplicate ID |
| 21:09 | <Hixie> | zcorpan_: i got a 500 error trying to reply to that |
| 21:09 | <Hixie> | wonder why |
| 21:09 | <Hixie> | oh wait |
| 21:10 | <Hixie> | my entire site is giving issues |
| 21:10 | Hixie | gets on it |
| 21:10 | <Hixie> | uh oh |
| 21:10 | <Hixie> | i'm guessing acid3 hit some site |
| 21:10 | <zcorpan_> | css3.info |
| 21:10 | <Hixie> | some... bigger site |
| 21:11 | <zcorpan_> | css3.info claims the test is completed now |
| 21:12 | eseidel | thinks it's slightly scary that Hixie moved it to the main url :) |
| 21:13 | <Hixie> | is there a unix tool like uniq that will count how many instances of a line there were? |
| 21:13 | <Hixie> | oh nevermind |
| 21:14 | <Hixie> | uniq will do it |
| 21:14 | <Hixie> | -c |
| 21:14 | <Hixie> | http://www.cdr.cz/a/23477 seems to be the source of most hits |
| 21:14 | <Hixie> | hahahahahaha |
| 21:14 | <Hixie> | netscape 4 just crashes on acid3 |
| 21:15 | <Hixie> | good times |
| 21:16 | <Hixie> | http://www.genbeta.com/2008/01/31-acid3-test-nuevos-navegadores-nuevas-pruebas also drove a lot of traffic |
| 21:16 | <Hixie> | and http://zaistniejwsieci.pl/2008/01/30/acid3-gotowy-przetestuj-swoja-przegladarke/ |
| 21:16 | <Hixie> | all three drove hundreds more than css3.info |
| 21:16 | <gsnedders> | Eastern Europe, interesting. |
| 21:18 | <eseidel> | I guess adding Analytics to the test itself would kinda defeat the purpose :) |
| 21:18 | <Hixie> | :-) |
| 21:21 | <gsnedders> | eseidel: just use really complex ECMAScript for it! :P |
| 21:21 | <gsnedders> | then claim nobody visits it, because nothing ever registers as visiting it. |
| 21:22 | <Philip`> | IE 5.0 just has a script error and says "Scripting must be enabled to use this test." |
| 21:23 | <Philip`> | IE 5.5 gets "6 100" |
| 21:23 | <SadEagle> | konq 3.5.8 crashes :( |
| 21:23 | <Philip`> | Oh, it went up to "14 100" |
| 21:23 | <Hixie> | zcreplied |
| 21:23 | <Hixie> | er |
| 21:23 | <Hixie> | zcorpan left. |
| 21:23 | <Hixie> | oh well. |
| 21:25 | <gsnedders> | oh, and on a totally unrelated note, any suggestions for learning C++? |
| 21:25 | <Hixie> | fix some webkit bugs? :-) |
| 21:25 | <gsnedders> | Hixie: that's hard when you know no C++ :) |
| 21:25 | <Hixie> | probably a good way to learn :-) |
| 21:25 | <Hixie> | (i'm serious) |
| 21:26 | <gsnedders> | need some to start with, really, though. |
| 21:26 | <SadEagle> | gsnedders: what do you know now? |
| 21:26 | <gsnedders> | SadEagle: PHP, Python, and a little Obj-C |
| 21:26 | <gsnedders> | (but when I say little I mean really little) |
| 21:27 | <gsnedders> | SadEagle: oh, and I know BASIC from school |
| 21:27 | <othermaciej> | gsnedders: in all seriousness, it's a pretty good way to learn, if you already know some programming |
| 21:27 | gsnedders | spits at the ground at the very mention of BASIC |
| 21:28 | <Philip`> | Why do you want to learn C++? |
| 21:28 | <othermaciej> | gsnedders: though doing a few tutorial excercises in a language with explicit pointers might help preparation |
| 21:28 | <gsnedders> | Philip`: ironically for this discussion, WebKit |
| 21:28 | jgraham_ | predicted that answer |
| 21:29 | <SadEagle> | gsnedders: good. That's the value of learning BASIC: it teaches you what a bad/primitive programming language looks like, so you can appreciate the important language features it lacks :-) |
| 21:29 | SadEagle | kidnaps gsnedders |
| 21:29 | <Philip`> | gsnedders: You could learn by writing patches for Mozilla |
| 21:29 | <Hixie> | i'd definitely start on webkit rather than mozilla |
| 21:29 | <othermaciej> | gsnedders: pointers and memory management are the most likely things to trip up someone who's only programmed in higher-level languages |
| 21:30 | <gsnedders> | SadEagle: "jeff!!! was NOT here either, what have u done with him?! i hope u havent jeff-napped him or something, because beleive me, hes mine, FOR EVER!" (to quote a girl from school) |
| 21:30 | <Hixie> | unless what you're trying to learn is appreciation for sane code |
| 21:30 | <gsnedders> | (but she can't spell my name, and she calls me geoff, so ergh) |
| 21:31 | <gsnedders> | othermaciej: pointers I probably do more or less know enough about, it's the memory management where I'll need the most practise |
| 21:31 | <Dashiva> | Crazy++ |
| 21:31 | <gsnedders> | Hixie: I've got that from code I wrote now around three years ago that I still maintain, having never totally rewritten it :) |
| 21:31 | <Dashiva> | Memory management is simple, it's just an eternal process of not slipping up |
| 21:32 | <Hixie> | zcorpan_: replied to that forum post |
| 21:32 | <zcorpan_> | hsivonen: if i got the numbers right, about 27% used the html5 doctype in your sample |
| 21:32 | <gsnedders> | in C++, does anything free memory automagically? Leaving a function and local variables in that? |
| 21:32 | <Philip`> | Memory management is just pointers, and remembering what they're pointing at |
| 21:32 | <zcorpan_> | Hixie: thanks |
| 21:33 | <gsnedders> | othermaciej: as long as you can stop anyone from killing me for my useless bad buggy patches, I'll try sometime :) |
| 21:35 | <othermaciej> | gsnedders: there's stack memory and heap memory |
| 21:35 | <Philip`> | gsnedders: Whenever memory is allocated with "new", it has to be freed with "delete", and that's about the only rule; everything else is automatic |
| 21:35 | <othermaciej> | gsnedders: stack memory is freed when you call the function, but not heap memory |
| 21:35 | <othermaciej> | however, many C++ programs use a pattern where heap memory is pointed to by a stack object with a destructor that frees it automatically |
| 21:36 | <gsnedders> | I'm sure I'm more confused now than I was a minute ago. |
| 21:36 | <Dashiva> | Philip`: Unless you malloc :) |
| 21:37 | <othermaciej> | gsnedders: that's because it's confusing (tragically) |
| 21:37 | <zcorpan_> | hsivonen: ah, make that 29%... |
| 21:37 | <Dashiva> | They didn't invent gc because doing it manually was fun |
| 21:37 | <Philip`> | Dashiva: or calloc or realloc or strdup or I can't remember what else, but I was trying to avoid details :-p |
| 21:37 | <gsnedders> | how useful do you think it'd be to get back in practice with the little Obj-C I once knew before trying to take on C++> |
| 21:37 | <gsnedders> | s/>/?/ |
| 21:38 | <Philip`> | Just replace all occurrences of T* with std::tr1::shared_ptr<T*> and you'll be fine |
| 21:38 | gsnedders | squints |
| 21:38 | <Dashiva> | I still have nightmares about stl core code |
| 21:40 | <Philip`> | It might be useful to do something with plain C (write chtml5lib!) if you want to learn how manual memory management works so you can understand what C++ is doing |
| 21:41 | <gsnedders> | Philip`: learning C how? :P |
| 21:41 | <Philip`> | By writing code :-) |
| 21:41 | <gsnedders> | while (need_to_learn) learn_another_language(); |
| 21:41 | <othermaciej> | Philip`: now you're just trying to confuse him further |
| 21:41 | <Philip`> | #define need_to_learn 1 |
| 21:42 | <SadEagle> | Philip`: unfortunately, the first thing C programmers learning to write C++ need to do is to forget C. |
| 21:42 | <othermaciej> | gsnedders: Obj-C and C++ are not that deeply conceptually different |
| 21:42 | <othermaciej> | (though ObjC developers will tell you otherwise) |
| 21:43 | <gsnedders> | what's Safari itself written in, BTW? |
| 21:43 | <Philip`> | You could try OCaml instead |
| 21:43 | <Philip`> | That's got a C in its name |
| 21:43 | <gsnedders> | Philip`: LOL |
| 21:43 | <gsnedders> | really _useful_ for C++. :P |
| 21:43 | <jgraham_> | Incidentally if you do decide to contribute to chtml5lib , it might bubble further up my priority queue |
| 21:44 | gsnedders | ought to have a to-do list on backing-store and not just in memory |
| 21:44 | Philip` | has a to-do set rather than a to-do list |
| 21:44 | <jgraham_> | gsnedders: Well they start all the comp-sci's here on ML (which I think OCaml is based on) so it must be good for learning something |
| 21:45 | <gsnedders> | jgraham_: here == Cam, right? |
| 21:45 | <jgraham_> | Indeed |
| 21:45 | <jgraham_> | Philip` could tell you more :) |
| 21:45 | gsnedders | wonders if jgraham_ and Philip` will be around in mid-May, or if they'll have run away |
| 21:46 | jgraham_ | will be around mot of May, but probably not the 27-30th |
| 21:46 | <jgraham_> | s/mot/most/ |
| 21:46 | <Philip`> | The first term has ML, but the second has Java and most of the rest of the course is also Java :-( |
| 21:46 | gsnedders | will be in Cambridge around the 14th, most likely |
| 21:47 | <jgraham_> | gsnedders: Any reason |
| 21:47 | <jgraham_> | ? |
| 21:47 | <SadEagle> | ML is quite tighly connected with formal types. |
| 21:47 | Philip` | will probably be here nearly constantly for the next few years |
| 21:47 | <gsnedders> | jgraham_, Philip`: see you (mad) people :) |
| 21:48 | <jgraham_> | Philip`: All the people I know who did NatSci+25%CompSci 1A and then switched to NatSci still complain about the ML 7 years later :) |
| 21:49 | <jgraham_> | ("All the people" being 2, so not a great sample) |
| 21:50 | <hsivonen> | zcorpan_: yeah 29% |
| 21:52 | <gsnedders> | jgraham_, Philip`: But yeah, if either of you are around, I'm more than willing to meet up with either (or both) of you sometime while I'm there |
| 21:52 | <jgraham_> | gsnedders: Sure |
| 21:53 | gsnedders | doesn't know for certain when he'll be there |
| 21:53 | <jgraham_> | Is it me or is the HTMLWG pretty much inactive at the moment? |
| 21:54 | <gsnedders> | I'm going down for a memorial service for my grandmother, at an undecided time, who lived in Cambridge for c. 50 years until a few weeks before she died |
| 21:54 | <gsnedders> | jgraham_: yeah |
| 21:56 | <hsivonen> | jgraham_: let's see how much reaction my suggestion to make border='0' and friends conforming gets |
| 21:56 | gsnedders | runs off |
| 21:56 | gsnedders | waves |
| 21:58 | <Hixie> | jgraham_: lots of activity on the -comments list (relatively speakiing) |
| 21:58 | <jgraham_> | hsivonen: Well I didn't get a single response, positive or negative about table headers, which I found a little surprising. I don't expect that meant tacit agreement from all parties though :) |
| 21:58 | <Hixie> | but i wouldn't complain about the group being inactive... |
| 22:02 | <jgraham_> | Hixie: The only concern I have is that all the detailed discussions that people will expect to happen before CR aren't happening and so the spec will get stuck later in a done but unpublished state (though maybe that doesn't matter if vendors don't care) |
| 22:04 | <Hixie> | the longer the bikeshedding is delayed, the better, since once the spec is implemented, it's set in stone. |
| 22:04 | <hsivonen> | Hixie: I don't consider the header association algorithm bikeshedding |
| 22:05 | <Hixie> | sure, i was just talking about "the detailed discussions that people will expect to happen before CR" that aren't happening |
| 22:05 | <Hixie> | i assumed that meant bikeshedding :-) |
| 22:05 | <hsivonen> | ok |
| 22:05 | <hsivonen> | heh |
| 22:05 | <Hixie> | i'm not seeing any discussions that i want to have happen that aren't happening |
| 22:06 | <Hixie> | but then i'm still dealing with the discussions from 2 years ago |
| 22:06 | <hsivonen> | Hixie: what's your inital reaction to making border='0', language='JavaScript', etc. conforming? |
| 22:07 | <Hixie> | border=0 i'm against |
| 22:07 | <Hixie> | actually i guess i'm against both |
| 22:07 | <Hixie> | though language='JavaScript' is the default so it'd be harmless, i think |
| 22:07 | <Hixie> | what's the advantage, other than hiding those errors from users? |
| 22:07 | <Hixie> | er, authors |
| 22:07 | <hsivonen> | Hixie: border='0' is a cowpath if there ever was one |
| 22:08 | <Hixie> | border=0 is not the default |
| 22:08 | <zcorpan_> | it should be... :) |
| 22:08 | <Hixie> | that's a different matter |
| 22:08 | <hsivonen> | Hixie: hiding them leaves attention to stuff that matters |
| 22:08 | <zcorpan_> | it is in opera and safari |
| 22:08 | <jgraham_> | zcorpan_: Wrong tense :) |
| 22:08 | <Hixie> | hsivonen: if i was an author, i would want to know about spurious attributes like those |
| 22:08 | <SadEagle> | zcorpan_: it sort of used to be in konq, and people whined :-) |
| 22:09 | <zcorpan_> | SadEagle: really? |
| 22:09 | <hsivonen> | Hixie: depends a lot on whether updating existing site or syndicading content or whether creating a new bandwidth-optimized site |
| 22:09 | <SadEagle> | zcorpan_: well, not as much about border 0, but about a solid border and not a cheesy 3D effect, IIRC |
| 22:10 | <Philip`> | SadEagle: Indeed - subsets of ML are used here for teaching about types, and about functional programming and semantics and things, so I suppose it actually comes up nearly as much as Java (though far less in terms of time spent programming in the languages) |
| 22:10 | <Hixie> | hsivonen: i wouldn't object to a validator making a value judgement and moving all that kind of errors to a "site cleanup issues" section of the results |
| 22:10 | <jgraham_> | hsivonen: Also there is a marketing reason not to make too many presentation attributes conforming, because you loose but-in from the type of people who speak at web conferences and those people can have substantial effects on people's markup usage |
| 22:10 | <Hixie> | hsivonen: but i don't think we should make them allowed |
| 22:10 | <Hixie> | s/loose but-in/lose buy-in/ :-) |
| 22:10 | <Philip`> | gsnedders: Sounds like a plausible plan :-) |
| 22:10 | <jgraham_> | :-p |
| 22:10 | <Hixie> | though the loose butting-in of those people... |
| 22:10 | <hsivonen> | jgraham_: that's why only '0' or '100%' should be conforming |
| 22:11 | <hsivonen> | jgraham_: it doesn't allow full CSSless design |
| 22:11 | <Hixie> | i think that'd be too confusing a message |
| 22:11 | <Hixie> | "you can do this but only for these values" |
| 22:11 | <Hixie> | better to just have a clear message |
| 22:11 | <Hixie> | "no presentation stuff in html, use css" |
| 22:11 | <jgraham_> | hsivonen: HTML5 has too many weird quirks already |
| 22:11 | <Hixie> | yeah, really |
| 22:11 | <zcorpan_> | look what happened with <font> |
| 22:12 | <zcorpan_> | i think border='0' is in the same bucket |
| 22:12 | <zcorpan_> | to be honest |
| 22:13 | <Hixie> | what happened with <font>? |
| 22:13 | <zcorpan_> | people thought html5 was the worst idea ever because it allowed <font> |
| 22:14 | <Hixie> | people thought html5 was the worst idea ever because it didn't allow style="", too |
| 22:14 | <Hixie> | so i don't know |
| 22:14 | <Hixie> | but yes |
| 22:15 | <Hixie> | i agree with your conclusion |
| 22:15 | <zcorpan_> | lack of style='' didn't get the same attention as precense of font |
| 22:15 | <SadEagle> | did the lack of style thing change? |
| 22:15 | <jgraham_> | SadEagle: Not yet :) |
| 22:15 | <Hixie> | no, nor did the presence of <font> |
| 22:15 | <Hixie> | still not really sure how to deal with them |
| 22:15 | <SadEagle> | It seems some people replace style="" by a unique classname and such in the wild. |
| 22:15 | <zcorpan_> | in fact i haven't seen anyone complain about lack of style='' outside the wg |
| 22:15 | <jgraham_> | (validator.nu doesn't have an error or style="") |
| 22:16 | <jgraham_> | s/or/on/ |
| 22:16 | <Hixie> | i'm still considering having a "low quality" version of html5 conformance, which you can use in rapid prototyping |
| 22:16 | <SadEagle> | (which produces some honking ugly stylesheets) |
| 22:16 | <hsivonen> | jgraham_: It's too obvious that style='' needs to be allowed, so I never bothered taking it out |
| 22:17 | <jgraham_> | Hixie: I on't see a good reason to disallow style="" really. Simply disallowing style isn't going to magically make all pages work in a media independent way |
| 22:17 | <zcorpan_> | s/precense/presence/ |
| 22:17 | <Hixie> | jgraham_: nor is disallowing layout tables, but i still think we should do that |
| 22:18 | <Hixie> | style="" and .style have many use cases, but none actually apply to final-form web sites |
| 22:18 | <Hixie> | they all apply to prototyping, demos, media-specific code, etc |
| 22:18 | <zcorpan_> | .style is needed for animation stuff |
| 22:18 | <hsivonen> | Hixie: what about dbaron's point #2 in http://lists.w3.org/Archives/Public/www-style/2008Jan/0598.html ? |
| 22:19 | <Philip`> | Modern web applications are all about early releases and constant updates and perpetual betas, so things are always in prototyping and never reach a final-form state |
| 22:20 | <zcorpan_> | i've also used style='' for things that are on final-form web sites but for things that are there on a temporary basis because i don't want to mess with the external style sheet and its caching... |
| 22:20 | <jgraham_> | I thought some of the use cases put forward in the WG were convincing in the sense that there wasn't an obvious, equally painless method of achieving the effect without @style (though there might be a non-obvious or hard way of doing it) |
| 22:20 | <hsivonen> | I'd be interested to see Sam's take on border='0' considering how often it appears on planet intertwingly |
| 22:21 | <Hixie> | hsivonen: btw, content-type is in |
| 22:21 | <hsivonen> | Hixie: great. |
| 22:21 | <hsivonen> | Hixie: with mandatory text/html as the type? |
| 22:21 | <Hixie> | zcorpan_: no, you can use override style. but eventually, we should use safari's animation stuff, too. |
| 22:21 | <Hixie> | hsivonen: yes |
| 22:21 | <Hixie> | http://www.whatwg.org/specs/web-apps/current-work/#meta0 |
| 22:22 | <zcorpan_> | Hixie: override style? |
| 22:22 | <Hixie> | cssom stuff |
| 22:22 | <Dashiva> | But if .style is allowed and @style is not, you're left with something you can't serialize |
| 22:22 | <Hixie> | might not be implemented, though |
| 22:22 | <Hixie> | Dashiva: we already have that with <input> |
| 22:22 | <Hixie> | amongst other things |
| 22:23 | <hsivonen> | Hixie: "must have a value consisting either of the literal string "text/html;", followed by a single U+0020 SPACE character, followed by the literal string "charset=", followed by the character encoding name of the character encoding declaration." I can't parse "either" |
| 22:25 | <Hixie> | uh |
| 22:25 | <Hixie> | why is that word there |
| 22:25 | <Hixie> | ignore that word |
| 22:25 | <Hixie> | oh, copy/paste edit error |
| 22:26 | <hsivonen> | ok :-) |
| 22:26 | <Philip`> | Is content="text/html" not allowed? |
| 22:26 | <Hixie> | no |
| 22:26 | <Hixie> | not allowed |
| 22:27 | <zcorpan_> | can't that be abbreviated to "must have a value consisting of the literal string "text/html; charset=" followed by..." ? |
| 22:28 | <Hixie> | it could, except that it might word-wrap at the space |
| 22:28 | <Hixie> | and if i use a NBSP instead of a space, then people will get confused... |
| 22:28 | <Hixie> | and all kinds of problems like that |
| 22:28 | <hsivonen> | <nobr>! |
| 22:28 | <zcorpan_> | <nobr> ;) |
| 22:28 | <zcorpan_> | bah you beat me to it |
| 22:28 | <Dashiva> | white-space: nowrap; |
| 22:34 | <zcorpan_> | Hixie: i don't think the way html5 defines http-equiv reflects how browsers have to implement it |
| 22:34 | <Hixie> | hsivonen: |
| 22:34 | <Hixie> | 0020 / 400Bad value “_top” for attribute “target” on element “a” from namespace “http://www.w3.org/1999/xhtml”: Bad browsing context name: Browsing context name started with the underscore and used a reserved keyword “top”. |
| 22:34 | <Hixie> | hsivonen: ^ that seems like a bogus error (_top is allowed) |
| 22:34 | <Hixie> | zcorpan_: oh? |
| 22:35 | <zcorpan_> | Hixie: e.g., i think we have to support things like cache-control as meta, and refresh as a http header |
| 22:35 | <Hixie> | as defined, nothing stops you from handling refresh as an http header |
| 22:35 | <Hixie> | and if needed, we could add cache-control |
| 22:36 | <zcorpan_> | ok |
| 22:36 | <hsivonen> | Hixie: I'm pretty sure it wasn't allowe two hours ago |
| 22:36 | <zcorpan_> | i don't have data at hand but i'm pretty sure we'd get lots of bug reports if we dropped support for most http-equiv values |
| 22:36 | <Hixie> | hsivonen: i haven't changed that section of the spec this year :-) |
| 22:36 | <hsivonen> | whoa! |
| 22:37 | <Hixie> | zcorpan_: it would be helpful to know which are supported and which should be in the spec |
| 22:38 | <Hixie> | zcorpan_: there are certainly many that we _don't_ want to support, and quite possible some that we really do that aren't on the list |
| 22:38 | <hsivonen> | Hixie: fix compiling. thanks |
| 22:38 | <zcorpan_> | Hixie: true |
| 22:39 | <Hixie> | "0018 / 400Entity reference was not terminated by a semicolon." is an error I'd be interested in mitigating |
| 22:39 | <Hixie> | not sure how easily we can do so |
| 22:39 | <Hixie> | "0094 / 400Text after “&” did not match an entity name." too |
| 22:39 | <hsivonen> | Hixie: actually, I don't feel a need to eliminate the former. I want to eleminate the latter. |
| 22:40 | <hsivonen> | but even more, I want to make border='0' conforming |
| 22:40 | <Hixie> | why? |
| 22:40 | <hsivonen> | it is harmless except for bandwidth cost |
| 22:40 | <Hixie> | it's as harmless as <center>, no? |
| 22:41 | <hsivonen> | and editing it out of every single image embedding snippet offered on the Web is a waste of time |
| 22:41 | <hsivonen> | Hixie: <center> is a bit different |
| 22:41 | <hsivonen> | Hixie: <center> doesn't fix an awful browser default and it isn't as common |
| 22:42 | <hsivonen> | Hixie: OTOH, <center> makes centering easy while CSS doesn't |
| 22:42 | zcorpan_ | still thinks the browser default should be fixed ;) |
| 22:43 | <Hixie> | zcorpan_: rendering section. send feedback. :-) |
| 22:43 | <hsivonen> | zcorpan_: I think the browser default should be fixed for the sake of coming generations of Web authors |
| 22:43 | <hsivonen> | zcorpan_: and border='0' should be allowed to stop bothering the current generation |
| 22:44 | <zcorpan_> | makes sense, but i'm afraid that allowing border='0' is going to backfire because the current generation doesn't want html5 to allow "presentational markup" |
| 22:45 | <Hixie> | hsivonen: why would you rather the spec allow it as opposed to having the validator move that kind of error to a separate "harmless issues that you might want to fix to save bandwidth" area? |
| 22:46 | <hsivonen> | zcorpan_: the part of the current generation making top sites seems to be in habit of using it |
| 22:46 | <Philip`> | <meta name="content-type" content="..."> seems reasonably common - do browsers support that? |
| 22:47 | <hsivonen> | Hixie: I'd rather have validator warnings about wasteful but otherwise harmless stuff |
| 22:47 | <Hixie> | hsivonen: well, i guess i could be convinced that this is similar to the style="" attribute, and could be convinced that it would be something to put into a "low quality" conformance level |
| 22:48 | <hsivonen> | Hixie: I'm not a fan of conformance levels, so I'd prefer to bargain with errors |
| 22:48 | <hsivonen> | doh |
| 22:48 | <hsivonen> | warnigns |
| 22:48 | <hsivonen> | typo++ |
| 22:49 | <Hixie> | i don't think the spec should allow useless stuff |
| 22:49 | <Hixie> | harmless though it may be |
| 22:49 | <hsivonen> | Hixie: we are already down that slippery slope with /> and xmlns |
| 22:49 | <hsivonen> | (allowing those is good, imo) |
| 22:50 | <Hixie> | i'm against the whole xmlns="" nonsense and /> nonsense that we have now, but at least that has some semi-sane value (in terms of making it harder to argue that a document is xml and not html when sent as text/html) |
| 22:50 | <zcorpan_> | hsivonen: i thought you were against /> at first? |
| 22:50 | <Hixie> | wow, you have 4 instances of "Attribute “abbr” not allowed on element “th”" |
| 22:50 | <hsivonen> | zcorpan_: could be. I'm not any longer |
| 22:50 | <zcorpan_> | hsivonen: ok |
| 22:51 | <Hixie> | hsivonen: "Attribute "accept-charset" not allowed on element "form" at this point." seems bogus, isn't accept-charset allowed on <form>? |
| 22:51 | <Philip`> | http://philip.html5.org/data/meta-content-type.txt |
| 22:51 | <Philip`> | Hmm, content="text/html" isn't that common |
| 22:53 | <zcorpan_> | Philip`: hmm. opera thinks that's application/octet-stream |
| 22:53 | <zcorpan_> | content sniffing++ |
| 22:53 | <gsnedders> | Saf3 too |
| 22:53 | <hsivonen> | Hixie: fix in build |
| 22:53 | <Hixie> | Philip`: i've made the space in the content-type stuff optional, given that data |
| 22:54 | <hsivonen> | Hixie: given Philip`'s data, the space should probably be optional |
| 22:54 | <Hixie> | hah |
| 22:54 | <Hixie> | way ahead of you |
| 22:54 | <Hixie> | :-P |
| 22:54 | <hsivonen> | thanks |
| 22:54 | <Philip`> | zcorpan_: Opera 9.2 says "Type text/plain" after popping up the file-download box |
| 22:54 | <Hixie> | I assume that "Attribute "alt" not allowed on element "input" at this point." was because the type="" attribute wasn't "image" |
| 22:55 | <Philip`> | so at least it's not lying about the type, but it is handling it wrong |
| 22:55 | <Philip`> | (for some definition of "wrong") |
| 22:55 | <zcorpan_> | Philip`: i pressed open, then it showed it as normal text/plain but the info sidebar says it's a/o-s |
| 22:55 | <hsivonen> | Hixie: yeah |
| 22:56 | <Hixie> | hsivonen: another reason i'm against border=0 being allowed is that i don't want to have anything deprecated. if we're trying to phase something out, it shouldn't be allowed. that's why i was against adding the meta http-equiv=content/type thing, but eventually i just realised we shouldn't bother trying to get rid of it, and that's why i added it -- i don't treat it as something we're phasing out. |
| 22:56 | <Hixie> | hsivonen: i really do want to phase <img border=0> out |
| 22:57 | <Hixie> | same with <script language> |
| 22:57 | <hsivonen> | Hixie: I think we don't need to deprecate stuff that we don't particularly encourage authors to use |
| 22:57 | <Hixie> | deprecate = not particularly encourage |
| 22:58 | <hsivonen> | Hixie: we can just allow those, put in warnings and let them fade without deprecation |
| 22:58 | <hsivonen> | deprecate = source of endless debates |
| 22:58 | <Hixie> | if we want them to fade, then we shouldn't allow them, imho |
| 22:58 | <Hixie> | i think growing pains type stuff belongs in the validator as a ui distinction, not in the spec |
| 22:59 | <hsivonen> | I can let them fade, but I don't particularly want them to |
| 22:59 | <Hixie> | i do :-) |
| 23:00 | <hsivonen> | Hixie: I already have the Lax Content-Type box for RFC 3023... |
| 23:00 | <Philip`> | jruderman: It's easy to find ~1000 https sites from dmoz.org, to look for invalid certificates, but I don't if that's enough to be informative |
| 23:01 | <hsivonen> | Hixie: I think needing a checkbox for turning a spec off is a sign of the spec needed adjustment |
| 23:02 | <hsivonen> | s/needed/needing/ |
| 23:02 | <Hixie> | hsivonen: i'm not suggesting a checkbox |
| 23:02 | <Hixie> | hsivonen: i'm suggesting that the report separate those errors out and list them after the other errors, as things that are errors but likely harmless. |
| 23:03 | <Hixie> | "Bad value "2008-01-18T14:33:01-0800" for attribute "datetime" on element "time"" is an interesting error, maybe we should relax that syntax, if that is found to be common. more study required. |
| 23:03 | <Philip`> | Hmm, Opera displays http://philip.html5.org/data/meta-content-type.txt as utf-8 but Firefox displays it as iso-8859-1 |
| 23:04 | <gsnedders> | Hixie: IMO we should allow all possible ISO8601 forms of calendar date + time |
| 23:04 | <gsnedders> | (but not other date forms like week dates and ordinal dates) |
| 23:04 | <Hixie> | that's a whole lot of syntax |
| 23:04 | <Hixie> | there's little chance of that happening |
| 23:05 | Philip` | adds an HTTP charset |
| 23:05 | <Philip`> | ...and now Opera seems to display it as text properly, rather than sniffing it as undisplayable |
| 23:06 | <gsnedders> | Hixie: I don't think we need to change that much, though |
| 23:07 | <gsnedders> | or rather, if we require it to be a full date/time format |
| 23:07 | <Hixie> | not sure what you mean |
| 23:08 | <gsnedders> | we actually require everything up to seconds to be present |
| 23:09 | <Hixie> | i think ISO8601 allows more than you realise |
| 23:09 | <gsnedders> | like what? |
| 23:09 | <Hixie> | W23-2004 |
| 23:09 | <Hixie> | and all kinds of other stuff |
| 23:09 | <Hixie> | it's a crazy spec that defines all kinds of random stuff |
| 23:10 | <gsnedders> | Hixie: "calendar date" as I said above excludes week dates like that |
| 23:10 | <gsnedders> | Hixie: "calendar dates expressed in terms of calendar year, calendar month and calendar day of the month |
| 23:10 | <gsnedders> | (to quote ISO8601:2004) |
| 23:11 | <Hixie> | you have access to the spec? |
| 23:12 | <gsnedders> | yes, I had to implement it once. Not fun. |
| 23:12 | <Hixie> | ah |
| 23:12 | <gsnedders> | Horrible spec, actually. |
| 23:12 | Hixie | doesn't recall ever actually seeing it, only having second hand reports of it |
| 23:13 | <gsnedders> | First and second editions apparently allow even more than the forth edition apparently |
| 23:14 | <hsivonen> | the used to be final committee drafts on the Web |
| 23:14 | <hsivonen> | s/the/there/ |
| 23:15 | <Hixie> | i should get google to just buy me a copy |
| 23:15 | <gsnedders> | <http://feedparser.org/docs/date-parsing.html>; lists some forms that aren't valid under ISO8601:2004 and only under earlier editions |
| 23:15 | <gsnedders> | s/and only/but are/ |
| 23:16 | <gsnedders> | the -yy ones are all invalid ISO8601:204 |
| 23:16 | <gsnedders> | *2004 |
| 23:18 | <Philip`> | It'll be interesting to see how date/time formats develop once humans start regularly moving relativistically |
| 23:20 | <jgraham_> | s/once/if/ :) |
| 23:21 | Hixie | has now read every error in henri's datafile |
| 23:23 | <Philip`> | http://philip.html5.org/data/meta-http-equiv.txt - "Содержимое-Тип"? |
| 23:23 | Philip` | guesses that's Russian for Content-Type |
| 23:24 | <Philip`> | http://tnt.udm.ru/ - ah, it is |
| 23:25 | <Hixie> | yay for curl | sort -n |
| 23:28 | <Philip`> | Oh, good idea |
| 23:28 | Philip` | updates the files so they're sorted by default |