| 12:43 | <hsivonen> | Hixie: should headers='' and summary='' be considered removed or not added yet? |
| 12:44 | <annevk> | when he did tables they were "removed" |
| 12:44 | <annevk> | well, not added because it wasn't clear what their processing model and use case was |
| 12:56 | <Lachy> | headers is useful for associating cells with their headers in complex tables. They're also more supported by ATs than scope="" |
| 12:56 | <Lachy> | but that could be fixed by defining the processing model for scope in a way that ATs can implement |
| 12:57 | <Lachy> | I think summary wasn't added few people use it, and even fewer use it correctly |
| 12:57 | <annevk> | I don't think much people use headers either |
| 12:58 | <annevk> | But I should shut up as I don't have data |
| 12:58 | <Lachy> | I've used headers="" before |
| 12:59 | <Lachy> | but I mostly use scope="" |
| 17:02 | <Lachy> | updated the FAQ with a new entry, thanks to zcorpan http://blog.whatwg.org/faq/#tag-soup |
| 17:04 | <zcorpan> | Lachy: your revised text is much better than my original :) |
| 17:04 | <Lachy> | :-) |
| 17:41 | <gsnedders> | on that subject… |
| 17:42 | <gsnedders> | annevk: does your XML5 have the same document conformance requirements as XML 1.0, so that any XML5 document can be parsed as XML 1.0? (seeming the real failing on XML 1.1 was the lack of compatiblity in a similar way) |
| 17:43 | <annevk> | no |
| 17:43 | <annevk> | that's by definition impossible since "XML5 document" will depend on the MIME type, not on the document |
| 17:43 | <Lachy> | but conforming XML 1.0 will be conforming XML5, so XML5 is a superset |
| 17:44 | <annevk> | yeah, although I'd like to simply forbid DOCTYPEs |
| 17:44 | <Lachy> | (unless you're going to make DOCTYPEs and DTDs non conforming) |
| 17:44 | <annevk> | :) |
| 17:44 | <gsnedders> | but I guess they won't cause a fatal error :P |
| 17:44 | <gsnedders> | </sarcasm> |
| 17:44 | <Lachy> | you could make external subsets non-conforming |
| 17:45 | <annevk> | external subsets are not supported |
| 17:45 | <Lachy> | yeah, right, so any document that uses one, like <!DOCTYPE foo SYSTEM "http://..."> will be non-conforming |
| 17:45 | <annevk> | currently it gives you a parse error after "<!DOCTYPE" |
| 17:46 | <Lachy> | so <!DOCTYPE html> will be non-conforming XML5? cool@ |
| 17:46 | <Lachy> | ! |
| 17:47 | <annevk> | i'm sure it will be changed in due course |
| 17:47 | <annevk> | but for the moment it's ok I think |
| 17:48 | zcorpan | thinks that conforming xml5 has to be the same or a subset of well-formed xml 1.0 in order to be successful, looking at xml 1.1, but i could be wrong |
| 17:48 | <zcorpan> | otherwise there will be a de-facto subset of xml5 that is also well-formed xml 1.0 |
| 17:48 | <zcorpan> | which might be ok |
| 17:49 | <zcorpan> | when xml 1.0 processors are not longer used that subset becomes irrelevant :) |
| 17:49 | <annevk> | If people start relying on my single new feature </> XML 1.0 processors might be encouraged to update |
| 17:49 | <Lachy> | the major problem with XML 1.1 was that it wasn't even backwards compatible. There was no subset of 1.1 that was conforming 1.0 |
| 17:49 | <gsnedders> | close last open tag? |
| 17:49 | <annevk> | CloseTagShort |
| 17:49 | <annevk> | yes |
| 17:49 | <zcorpan> | ah, i like </> :) |
| 17:50 | <hasather> | it should've been in XML from the beginning IMO |
| 17:50 | <gsnedders> | doesn't SGML have something similar? |
| 17:50 | <annevk> | This is all highly theoretical of course because I'm not sure how much people like it. |
| 17:50 | <gsnedders> | like, exactly that? |
| 17:50 | <hasather> | it was discussed |
| 17:50 | <Lachy> | nice! Defining failed SGML features is sure to work! :-) |
| 17:50 | <zcorpan> | gavin: yes |
| 17:50 | <zcorpan> | annevk: can't you make end tags optional too? |
| 17:51 | <annevk> | I'm not sure whether that's conforming or a parse error |
| 17:51 | <Lachy> | omitting end tags should be a parse error |
| 17:51 | <annevk> | The idea is that <a><b></a> will give you <a><b></b></a> as tree |
| 17:52 | <annevk> | The idea is also for that to be non-conforming markup |
| 17:52 | <Dashiva> | Is there a WD on XML5 somewhere? |
| 17:52 | <Lachy> | consider the case where someone writes <a>foo <b>bar<b> baz</a>. They probably meant </b>, but mistyped, so it would be good for conformance checkers to emit an error |
| 17:53 | <annevk> | Dashiva, I'm implementing first |
| 17:53 | <Lachy> | other than just allowing more Unicode chars, and the possible </>, what other features will be incompat with 1.0? |
| 17:53 | <annevk> | http://annevankesteren.nl/2007/xml5lib-tokenizer-plan.txt has an incomplete tokenization plan though, fwiw |
| 17:53 | <annevk> | Lachy, <x foo bar=x /> |
| 17:54 | <Lachy> | why are you making that conforming? |
| 17:54 | <annevk> | for timbl |
| 17:54 | <Lachy> | ? |
| 17:55 | <annevk> | that was a joke sort of; mainly for compat with HTML |
| 17:55 | <Lachy> | I don't think it should be made conforming, though I agree with making the error handling compat with HTML in that case |
| 17:55 | <zcorpan> | annevk: won't </> being conforming as an end tag in xml5 make people try to use it in html? (cf <foo />) |
| 17:56 | <annevk> | zcorpan, maybe |
| 17:58 | <Lachy> | I'd advise against making too many incompat changes from XML 1.0. It's going to be hard enough getting people XML fans to accept XML5's error handling, without making it seem even more like tag soup |
| 17:59 | <zcorpan> | annevk: have you researched what non-drocanian xml parsers do? feed parsers, for isntance? |
| 17:59 | <Lachy> | just look at how many arguments there have been against HTML defining error handling! I wish you luck with XML ;-) |
| 17:59 | <zcorpan> | yeah |
| 17:59 | gsnedders | shudders |
| 17:59 | gsnedders | then moves on the implement it regardles |
| 17:59 | <gsnedders> | *regardless |
| 18:00 | <annevk> | zcorpan, a little bit, not much |
| 18:00 | <zcorpan> | ok |
| 18:00 | <annevk> | zcorpan, I don't think those non-draconian parsers do the interesting bits, such as DOCTYPEs |
| 18:00 | <zcorpan> | e.g., how they handle </> might be relevant |
| 18:01 | <annevk> | ah, we just put a WHATWG sticker on it and ship it |
| 18:01 | <annevk> | zcorpan, how easy they seem to be willing to switch from one draconian parser to another suggests otherwise |
| 18:02 | <zcorpan> | i don't follow |
| 18:02 | <annevk> | Sam Ruby made a liberal XML parser based on html5lib and used that instead |
| 18:02 | <zcorpan> | right |
| 18:02 | <gsnedders> | it isn't overly imporatnt |
| 18:02 | <gsnedders> | they're very varied |
| 18:02 | <annevk> | as opposed to one based on an SGML parser |
| 18:03 | <annevk> | anyway, if something proves to be an issue we should of course change it |
| 18:03 | <gsnedders> | so reverse engineering them is pointless. all do different things really. |
| 18:03 | <zcorpan> | my point is that if a number of non-drocanian xml parsers treat </> differently from the way you planned it should work then it might not be a good idea to spec something different from what they already do |
| 18:03 | <zcorpan> | i don't know what they do, i'm just suggesting that it's something to look into |
| 18:04 | <zcorpan> | if all are very different then content probably doesn't depend on it :) |
| 18:04 | <annevk> | it is very likely they do something else |
| 18:04 | <annevk> | i think it's unlikely "XML content" will depend greatly on it though |
| 18:05 | <Lachy> | good night everyone |
| 18:05 | <annevk> | bye |
| 18:32 | <zcorpan> | jdandrea: made any progress on the style sheet thing? |
| 18:40 | <annevk> | Ok, the tokenizer now handles cases like <a x y>TEST<![CDATA[test]]>x</a> |
| 18:40 | <annevk> | And it actually works. I was planning to do entities to today, but it seems like it has to wait |
| 18:56 | <Dashiva> | I am bit depressed by how much of the tokenizer is just handling doctypes |
| 18:58 | <Dashiva> | 68 states, 42 are for doctype. Simplicity cries. |
| 19:28 | <jdandrea> | zcorpan: not yet (some multitasking afoot) but I did run through Hixie's excellent article on spec parsing (a few times, slowly). I expect to try a section today. To recap, I'm looking for author-specific normative statements (correct me if I'm wrong)? |
| 19:29 | jdandrea | is tending to an injured family member (they'll be fine) |
| 19:30 | <gsnedders> | jdandrea: what article? |
| 19:32 | <jdandrea> | gsnedders: http://ln.hixie.ch/?start=1140242962&count=1 - I've read RFC2119 and all but this was a really good review. Should be required reading by everyone on the HTML WG list IMNSHO. :) |
| 19:32 | <gsnedders> | jdandrea: ah. that old article. |
| 19:33 | <jdandrea> | an oldie but goodie. ;) |
| 20:28 | <zcorpan> | jdandrea: you should be looking for things that *don't* apply to authors :) |
| 20:31 | <zcorpan> | jdandrea: things that are author-specific but not normative should still be visible (e.g. examples) |
| 20:32 | <zcorpan> | jdandrea: and things that are ua-specific but non-normative should probably still be hidden (e.g. examples for how some algorithm works) |
| 21:23 | <gsnedders> | Hixie: you able to get a list of @rel values on HTML docs? |
| 22:33 | <met_> | shoudn't be first paragraph actualized? http://blog.whatwg.org/faq/#schedule |
| 23:03 | <Hixie> | "actiualised"? |
| 23:04 | <Hixie> | gsnedders: how do you mean? |
| 23:06 | <met_> | "When will HTML 5 be finished? - Around 15 years or more to reach a W3C recommendation" sound little bit sarcastic now |
| 23:08 | <Hixie> | why? |
| 23:08 | <Hixie> | it's about right |
| 23:08 | <met_> | if html wg name his next specification HTML5? |
| 23:08 | <Hixie> | yes, that's what the question is talking about |
| 23:09 | <met_> | there is schelude for 2010, isn't? |
| 23:09 | <Hixie> | haha |
| 23:09 | <Hixie> | yes |
| 23:09 | <Hixie> | but that schedule is a joke |
| 23:09 | <met_> | really? 8-) |
| 23:09 | <Hixie> | consider that work on HTML4 started around 1996, and still today, 11 years later, hasn't got two complete interoperable implementations |
| 23:09 | <met_> | but we need some specificaition earlier then in 2022 8-0 |
| 23:12 | <Hixie> | we have one today |
| 23:12 | <Hixie> | read the rest of the answer :-) |
| 23:14 | <met_> | reading "For a spec to become a REC, it requires two 100% complete and fully interoperable implementations", really it worked such way in W3C? im not sure |
| 23:14 | <met_> | not speaking it isn't good idea, of course it is |
| 23:16 | <met_> | "different parts of the specification are at different maturity levels" - it's ok in WHATWG, but surely it is not way how it will work in HTML WG? |
| 23:16 | <Hixie> | you'd have to ask DanC about how it'll work in the W3C |
| 23:16 | <Hixie> | gotta go |
| 23:16 | <Hixie> | bbl |
| 23:28 | <gsnedders> | Hixie: what values people use in the rel attribute |