| 00:16 | <Hixie> | ok |
| 00:16 | <Hixie> | <blockquote> is clearly not a sectioning element |
| 00:18 | <Hixie> | and outlines will not include headers in tables, figures, blockquotes, or datagrids. |
| 01:27 | <Hixie> | oh lord, my legacy lives on http://oxine.opera.com/documentation/dom-interface.html |
| 01:42 | <kingryan> | long live legacy! |
| 01:43 | <Philip`> | whether you want it to or not |
| 02:24 | <Hixie> | hsivonen: ok. so. tell me if the new algorithm is better. |
| 02:24 | <Hixie> | jgraham_: your input would be especially interesting too, since you implemented the last version :-) |
| 02:46 | <annevk> | Hixie, do you have a rough plan of what you're going to edit in the spec? |
| 02:48 | <Hixie> | how do you mean? |
| 02:50 | <annevk> | If you had some plan which sections of the specification you're going to work on for the next month or so. I guess I'm mostly interested in when the parsing section gets another update. |
| 02:50 | <Hixie> | no particular plan. |
| 02:50 | <Hixie> | i can start on parsing next if you want. |
| 02:53 | <annevk> | Cool. In particular I'm interested in seeing insertion modes being replaced with phases and updates to DOCTYPE sniffing. |
| 02:53 | <Hixie> | yeah |
| 02:53 | <Hixie> | that scares me |
| 06:04 | <takkaria> | Hixie: how do you twittr the svn commits with whatwg? I want to copy it for one of my projects |
| 06:56 | <Hixie> | takkaria: send me mail, i'll send you the script |
| 07:05 | <takkaria> | Hixie: ta |
| 07:07 | <takkaria> | sent |
| 07:13 | <Hixie> | ok, sections are updated. e-mail sent. |
| 08:42 | <Lachy> | Can anyone explain the logic behind the suggestion to use <dl> for marking up poetry? Even the wiki page with the example seems to have no real explanation |
| 08:47 | <virtuelv> | seems bizarre |
| 08:48 | <Lachy> | it seems that some people, when determining what kind of elements are "prose content" just look at those that say "prose content" in the category list, and fail to look at the definition of prose content |
| 08:48 | <virtuelv> | IMO, poetry is a bad match for examples |
| 08:48 | <Lachy> | the heirarchy of categories should be made clearer somehow |
| 08:48 | <virtuelv> | there is poetry where whitespace is extremely significant |
| 08:48 | <Lachy> | for that, we have <pre> |
| 08:49 | <virtuelv> | yes |
| 08:49 | <virtuelv> | my point is just that using a poem in examples, or suggesting anything particular for poems is bound to fail and cause misunderstandings |
| 09:39 | <Hixie> | prose content is a terrible name |
| 09:39 | <Hixie> | i really should fine something better |
| 09:41 | <Hixie> | also "grouping content" and "text-level semantics" are terrible section names |
| 09:59 | <jgraham_> | Hixie: I'll try to look at the heading outline stuff in a bit of detail soon |
| 09:59 | jgraham_ | is, in general, quite busy at the moment |
| 09:59 | <Hixie> | cool |
| 10:00 | <Hixie> | would be good to compare the new algorithm with the old one |
| 10:00 | <Hixie> | see if the same outlines come out |
| 10:00 | <Hixie> | they basically should |
| 10:00 | <Hixie> | though iirc there are some minor intentional differences, in edge cases with invalid markup in particular |
| 10:01 | <Hixie> | also <address> elements will get associated with anonymous sections now, not their DOM parent node <section> if that is a different section |
| 10:01 | <jgraham_> | Also, I'm glad you're still optimistic about FF3 having sane behaviour for unknown elements :) |
| 10:02 | <Hixie> | well they keep saying they'll try something |
| 10:02 | <Hixie> | they haven't given up yet |
| 10:03 | <Hixie> | we _really_ need new names for "prose content", "grouping content", and "text-level semantics" |
| 10:05 | <Hixie> | Maybe Flements, Bloments, and Inments. |
| 10:05 | <Hixie> | that would at least not confuse people into thinking they meant something else |
| 10:07 | <Hixie> | or Flodes, Grements, and Liliments |
| 10:07 | <annevk> | alpha beta and gamma |
| 10:08 | <Hixie> | the advantage of a made up word is that it fits well in the flow of text |
| 10:08 | <Hixie> | "<p> is a flode element" |
| 10:08 | <Hixie> | maybe "prose content" should be "flow content" |
| 10:09 | <Philip`> | You could have a <p> start flag |
| 10:09 | <Lachy> | flow content would work, since that's bascially the equivalent in HTML4 |
| 10:10 | <jgraham_> | I suggest not using flow content for the same reason :) |
| 10:13 | <Hixie> | ok flow content it is |
| 10:14 | <annevk> | did someone look at how XHTML2 solved this? |
| 10:14 | <jgraham_> | Does it exactly match the HTML 4 definition? |
| 10:15 | <annevk> | XHTML2 has modules |
| 10:15 | <annevk> | "XHTML Text Module" |
| 10:15 | <Hixie> | you're assuming xhtml2 solved it |
| 10:16 | <annevk> | s/solved/did/ |
| 10:16 | <Hixie> | well, they don't have the same concepts we do |
| 10:16 | <Hixie> | they don't have flow < phrasing < embedded, e.g. |
| 10:39 | <hsivonen> | hmm. it would be great to have a script that'd file a v.nu bug every time the spec svn commit message is marked as conformance-checker-relevant |
| 10:40 | <annevk> | at some point the tracker should support Atom with categories or something which might make that easy |
| 15:42 | <zcorpan> | Hixie: making <td> a sectioning root will make the outline algorithm useless for real-world pages where tables are used for layout |
| 15:43 | <zcorpan> | (unless it can be determindated which tables are used for layout and let those be ignored from the rule) |
| 15:47 | <zcorpan> | (however, i don't know when you'd have sections in <td> unless you were using a layout table, so perhaps <td> simply should not be a sectioning root) |
| 16:03 | <zcorpan> | Hixie: while you're renaming stuff, i think "text-level content" makes more sense than "phrasing content" |
| 16:07 | <zcorpan> | Hixie: html4 described inline as "text-level", and html4 has "phrase elements" (em, strong, dfn, etc) |
| 16:11 | <gsnedders> | Hixie: why do we have "current outlinee" with two "e"s? |
| 16:24 | <gsnedders> | Hixie: more nitpicking: steps one and two of creating an outline are wrong: they don't hold it yet |
| 17:06 | <annevk> | hmm, ES4 uses a namespace with a version number in it... |
| 17:32 | Philip` | sees that 15% of Alexa Top 500 pages are in HTML5 non-quirks mode, compared to 5% of dmoz.org pages |
| 17:33 | <Philip`> | (and 47% limited-quirks, vs 23%) |
| 17:33 | <gsnedders> | Philip`: and strict v. non-strict? |
| 17:34 | <Philip`> | (and 37% quirks, vs 72%) |
| 17:34 | <Philip`> | (and I hope those numbers add up right) |
| 17:35 | <Camaban> | I imagine dmoz has a lot more 'older' sites |
| 17:36 | <Philip`> | gsnedders: I'm not sure how to measure that exactly |
| 17:39 | <Philip`> | though if I simply count the top XHTML1.0-Strict and HTML4-Strict and XHTML1.1 doctypes, then that's 14% of Alexa and 4% of dmoz |
| 18:02 | <Philip`> | http://philip.html5.org/data/doctypes.html | http://philip.html5.org/data/doctypes-alexa.html |
| 18:05 | <Philip`> | http://philip.html5.org/data/doctypes-alexa.html#%3c%21doctype_html_public_%22-%2f%2fw3c%2f%2fdtd_html_4.01%2f%2fen%22_%22http%3a%2f%2fwww.w3.org%2ftr%2fhtml4%2fstrict.dtd%22_%2f%3e - HTML5 disagrees with Firefox and IE, which maybe isn't good |
| 18:06 | <Philip`> | (and Opera) |
| 18:11 | <zcorpan> | Philip`: thanks for providing links to the pages that have a given doctype |
| 18:12 | <Philip`> | zcorpan: I hope it doesn't matter that I skipped links for the more popular ones - they just made the page unreasonably huge |
| 18:12 | <Philip`> | (It's currently ~1.5MB, or 150KB with gzip, which is not entirely as small as one might wish) |
| 18:13 | <zcorpan> | Philip`: that's fine, i wanted to analyse the less common doctypes :) |
| 18:13 | <zcorpan> | since we're already more aligned with html5 than with ie on the commonly used doctypes |
| 19:08 | <Hixie> | zcorpan: yes, that was the entire point of making <td> a sectioning root. Also, I couldn't work out what order to make the headers go in, and I figured that in a _conforming_ case of a table with subsections (e.g. a character sheet where one of the cells is a character backstory), you wouldn't actually want them on the main outline. |
| 19:09 | <Hixie> | zcorpan: phrasing content can't be text-level content because phrasing content includes non-text things like <img> and <video>. |
| 19:09 | <Hixie> | gsnedders: "current outlinee" is correct (it's the thing being outlined) |
| 19:26 | <zcorpan> | Hixie: text-level means that it's on the same level as text, not that it is text |
| 19:26 | <Hixie> | yeah but i think that's confusing |
| 19:27 | <zcorpan> | i think phrasing is more confusing :) |
| 19:27 | <Hixie> | though possibly no more confusing than what we have now, indeed |
| 19:36 | <zcorpan> | Hixie: did you mean that the point of making <td> section root was to make the algorithm useless for real-world pages? |
| 19:37 | <Hixie> | well maybe saying it was the whole point was overstating the case a bit |
| 19:37 | <zcorpan> | ok |
| 19:37 | <Hixie> | on a compliant page that happens to use headers in tables, you'd not want those headers in the outline |
| 19:37 | <Hixie> | just like headers in a figure or in a blockquote, they are like a "subdocument" |
| 19:38 | <Hixie> | now, it does mean that outlines don't work in abusive pages, but i'm not shedding any tears over this |
| 19:38 | <zcorpan> | implementors that make outlines probably already know how to spot a layout table, so they could easily exclude those tables |
| 19:40 | <Hixie> | yeah |
| 19:40 | <Hixie> | maybe a quirks mode thing :-) |
| 19:40 | <zcorpan> | tables are used for layout in standards mode too in the wild |
| 19:40 | <Hixie> | yeah |
| 19:50 | <gsnedders> | Hixie: peh. that's confusing. |
| 19:51 | <Hixie> | significantly better ideas welcome :-) |
| 19:54 | <gsnedders> | "current thing being outlined" :P |
| 20:01 | <zcorpan> | Hixie: "authors are encouraged to place at most one top-level heading in each sectioning element" or some such... but perhaps that's not accurate enough or not understandable |
| 20:06 | <Hixie> | yeah |
| 20:06 | <Hixie> | but then again, that's what the current encouragemenet basically proposes |
| 20:11 | <zcorpan> | that's true |
| 20:42 | <gsnedders> | Hixie: what's a character not defined by unicode? from the document conformance section, it seems that not all non-characters are |
| 20:42 | <Hixie> | yeah, but some of them are defined to be permanently undefined |
| 20:42 | <Hixie> | whereas others are merely not yet defined |
| 20:43 | <gsnedders> | oh, wait. |
| 20:43 | <gsnedders> | I realise what I'm mistaking. |
| 20:43 | <Hixie> | i'm so sorry for anne. he's somehow ended up editing a spec with multiple phone calls per week trying to tell him what the spec should say |
| 20:43 | <Hixie> | sure am glad i dumped xmlhttprequest now though :-D |
| 20:44 | <gsnedders> | xxFFFE is the only code point that's always a non-character, as FDD0 to FDEF are only exactly that |
| 20:44 | <gsnedders> | that's what confused me |
| 20:44 | <Hixie> | eh? |
| 20:44 | <Hixie> | there are lots of non-character characters |
| 20:44 | <Hixie> | FFFF |
| 20:44 | <Hixie> | FFFE |
| 20:44 | <Hixie> | U+03FFFE |
| 20:44 | <Hixie> | etc |
| 20:44 | <Hixie> | see the list i just put in the spec, in fact :-) |
| 20:45 | <gsnedders> | I thought it was wrong, that's the only point :) |
| 20:45 | <Hixie> | ok :-) |
| 20:45 | <gsnedders> | but as I said, I'm being silly |
| 20:45 | <Hixie> | well i just copied it from xml 1.0 |
| 20:45 | <Hixie> | so... |
| 20:45 | <gsnedders> | I think FFFF is legal |
| 20:46 | <Hixie> | it's not |
| 20:46 | <gsnedders> | "Noncharacters consist of the values U+nFFFE and |
| 20:46 | <gsnedders> | U+nFFFF (where n is from 0 to 1016) and the values U+FDD0..U+FDEF. |
| 20:46 | <gsnedders> | " |
| 20:46 | <gsnedders> | OK, I'm wrong again (and so is my implementation, then) |
| 20:46 | gsnedders | realises his implementation is just too smart for himself |
| 20:46 | <gsnedders> | my implementation _is_ right. |
| 20:47 | <gsnedders> | `($codepoint & 0xFFFE) === 0xFFFE` does match FFFF |
| 20:48 | <Hixie> | or rather, it does match it |
| 20:48 | <Hixie> | but yes |
| 20:48 | <Hixie> | that's a fine check |
| 20:48 | <gsnedders> | If you aren't too tired to misread it :) |
| 20:48 | <gsnedders> | (that's from my PHP impl., FWIW) |
| 20:49 | <gsnedders> | Or rather, It only is if you aren't too tired to understand it |
| 20:49 | <gsnedders> | ergh. |
| 20:49 | <gsnedders> | I can't even do English :) |
| 20:49 | gsnedders | gives up, and shuts up |
| 20:54 | <Hixie> | knowing when to shut up is a good skill to have :-) |
| 20:54 | <gsnedders> | Hixie: I would rather the bit about when to throw a parse error to be clearer. Surrogates _are_ defined by Unicode (they just have no character assignments), for example |
| 20:54 | <Hixie> | there are no surrogate characters |
| 20:54 | <Hixie> | unless i'm misunderstanding you |
| 20:54 | <Hixie> | what is unclear about the spec? |
| 20:55 | <gsnedders> | ah. true. if you take it of the actual meaning of character, yeah. |
| 20:55 | gsnedders | was thinking of a character as being any code-point, but of course with non-characters that's dumb |
| 20:55 | gsnedders | shuts up, again |
| 20:56 | <Hixie> | U+.... is the character, not the value as it was in the original byte stream |
| 20:56 | <Hixie> | so if you are decoding as UTF-16, you can never end up seeing a U+.... character from the surrogate blocks |
| 20:56 | <gsnedders> | yeah, that's true |
| 20:56 | gsnedders | needs to wake up |
| 20:56 | <Hixie> | and if you do it as UTF-8, and you see one of those characters, it's not actually a surrogate character, it's a non-character |
| 20:56 | <gsnedders> | (or, alternatively, just go to bed) |
| 20:56 | <Hixie> | but anyway |
| 20:57 | <Hixie> | where is the part fo the spec that says not to put encoding declarations in the file? |
| 20:57 | Hixie | can't find it |
| 20:57 | gsnedders | didn't know the spec said that |
| 20:57 | <Hixie> | apparently people want me to remove it |
| 20:57 | <Hixie> | which i'm fine with |
| 20:57 | <Hixie> | but i can't find it... |
| 20:58 | <gsnedders> | latest revision to the spec could cause some documents to cause a heckuva lot of parse error, me thinks. but people manage to create a heckuva lot of parse errors anyway. |
| 20:59 | <Philip`> | Hixie: s/occurances/occurrences/ in a recent edit |
| 20:59 | <Hixie> | hate that word |
| 20:59 | <Hixie> | thanks |
| 21:00 | <kingryan> | Hixie: by "encoding declarations in the file", could people be referring to "<meta http-equiv="content-type" content="text/html; charset=utf-8" /> |
| 21:00 | <kingryan> | " ? |
| 21:00 | <Hixie> | kingryan: yes |
| 21:00 | <gsnedders> | on the subject of English, how is the spec both en-gb-x-hixie and en-us-x-hixie at once? |
| 21:00 | <Hixie> | gsnedders: it's mostly -us-, i just haven't fixed the declarations yet |
| 21:00 | <Hixie> | they're in a different file |
| 21:00 | <kingryan> | are people confusing that being allowed with that being required? |
| 21:01 | <Hixie> | kingryan: no, i'm pretty sure i once wrote that people should use Content-Type headers instead |
| 21:01 | <Hixie> | but i can't find it anymore |
| 21:02 | <Hixie> | well, can't find it |
| 21:02 | <Hixie> | oh well |
| 21:32 | <annevk> | woha, multimousewheel is gone again? |
| 21:32 | <annevk> | hmm |
| 21:32 | <annevk> | i wonder if they thought everything through, such as mousewheel not firing for certain types of scrolling currently |
| 21:33 | <Hixie> | i doubt it |
| 21:33 | <Hixie> | who's editing that spec? |
| 21:33 | <Hixie> | do they have any browser, qa, and spec review exerience? |
| 21:35 | <annevk> | dunno |
| 21:35 | <annevk> | well, the spec is edited by Andrew Emmons I believe |
| 21:35 | <Hixie> | don't know who that is |
| 21:36 | <annevk> | from BitFlash |
| 21:36 | <Hixie> | never heard of it |
| 21:36 | <annevk> | i don't know him either |
| 21:36 | <Hixie> | well, we'll see |
| 21:36 | <Hixie> | there really should be a way to train spec writers |
| 21:36 | <Hixie> | maybe i should write a book or something |
| 21:39 | <annevk> | "HOWTO BE A HIXIE" |
| 21:45 | <gsnedders> | 1) Start a cabal that everyone hates. |
| 21:45 | <gsnedders> | 2) ??? |
| 21:45 | <gsnedders> | 3) Profit! |
| 21:46 | <jgraham> | Spec writing for Dummies |
| 21:47 | gsnedders | realises in another virtual desktop he started a LaTeX file containing "\chapter{Evaluation}" hours ago. |
| 21:48 | gsnedders | has been doing well at procrastinating today |
| 21:48 | <gsnedders> | Philip`: I've just got an email from <mtanalin⊙yr> too |
| 21:49 | <jgraham> | gsnedders, Philip`: I get those |
| 21:50 | <jgraham> | Hixie: re the text "The outline for a sectioning content element or a sectioning root element consists of a list of one or more potentially nested sections. Each section can have zero or one heading associated with it. The algorithm for the outline also associates each node in the DOM tree with a particular section and potentially a heading." |
| 21:51 | <jgraham> | The "sections" referred to here are need not be actual <section> elements or anything, I assume. This could be more clear |
| 21:57 | <Hixie> | jgraham: clarified. |
| 21:57 | <Hixie> | (should be regenned in about 20 seconds) |
| 21:58 | gsnedders | needs less coursework and more spec-gen clone |
| 22:07 | <annevk> | gsnedders, now you make me go looking |
| 22:12 | <Hixie> | does anyone actually use ISO-8859-11 ? |
| 22:13 | <annevk> | that's off-topic |
| 22:14 | <Hixie> | well i have feedback here saying i should make -11 turn into win874 |
| 22:14 | <Hixie> | and i have no idea if it matters or not |
| 22:14 | <Hixie> | safari doesn't even support -11 |
| 22:14 | <Hixie> | (it does support 874) |
| 22:16 | <annevk> | given that 874 is supported and that other browsers do "support" -11 and that mapping is cheap... |
| 22:19 | <dbaron> | Hrm. Hixie's subject lines don't work very well when you have limited horizontal space for subjects. He's always writing "Re: [whatwg] several messages about" |
| 22:20 | <annevk> | "Make using a Win1252-specific byte when the document declared as ISO-8859-1 be a parse error." is that really worth it? |
| 22:20 | <dbaron> | Hixie, and if you're not aware, you should check mozilla/intl/uconv/src/charsetalias.properties for Mozilla's behavior |
| 22:20 | annevk | thinks ISO-8859-1 should be an alias for the former |
| 22:21 | <dbaron> | Mozilla seems to treat windows-874 and iso-8859-11 separately |
| 22:22 | <dbaron> | Hixie, er, actually, that's not the case |
| 22:22 | <dbaron> | but somebody decided to do that particular mapping in C++ instead |
| 22:22 | <dbaron> | so we treatt iso-8859-11 as windows-874, per bug 127755 |
| 22:31 | <Hixie> | dbaron: yeah |
| 22:32 | <Hixie> | annevk: adding more complications isn't cheap |
| 22:32 | <Hixie> | annevk: they add up |
| 22:32 | <annevk> | it's a simply hashtable entry |
| 22:33 | <annevk> | "iso-8859-1" : "windows-1252", "iso-8859-11" : "windows-874", ... |
| 22:33 | <Hixie> | no, it |
| 22:33 | <Hixie> | is far more than that |
| 22:33 | <Hixie> | not in code |
| 22:33 | <Hixie> | but in cost |
| 22:34 | <Hixie> | it further antagonises the tag, for instance |
| 22:34 | <Hixie> | it makes QA more complex |
| 22:34 | <Hixie> | it makes people who try to use ISO-8859-11 wonder why they're getting od results |
| 22:34 | <Hixie> | etc |
| 22:34 | <Hixie> | it makes people say the spec is complicated |
| 22:35 | <annevk> | ideally this would not be solved at the HTML5 level though |
| 22:35 | <annevk> | IANA already has synonyms for charsets |
| 22:35 | <annevk> | they could simply make iso-8859-1 a synomym and likewise for its friends |
| 22:36 | <Hixie> | well, if you can convince them of that, let me know |
| 22:37 | <annevk> | I don't think I'm old enough to deal with what seems to be a political mayhem |
| 22:38 | <annevk> | I guess at some point it might be worth trying to fix it... |
| 22:38 | <Hixie> | neither am i :-) |
| 22:39 | <Hixie> | gsnedders: so do i still need to deal with this doctype feedback or is anne's comment enough? |
| 22:39 | <annevk> | http://lists.w3.org/Archives/Public/public-forms/2008Feb/att-0080/2008-02-27.html#topic4 |
| 22:41 | Philip` | sees no iso-8859-1 in his 125K pages, and 61 windows-874s |
| 22:41 | <annevk> | iso-8859-1 or iso-8859-11 ? |
| 22:42 | <Philip`> | Uh |
| 22:42 | <Philip`> | -11 |
| 22:42 | <Hixie> | HAHAHAHAHAHAHAHA |
| 22:42 | <Hixie> | anne: boy i hope that was a minuting error |
| 22:43 | <Philip`> | (except as a substring in http://www.btfonsterteknik.com/ ) |
| 22:43 | <Hixie> | Philip`: cool, thanks |
| 22:43 | <Hixie> | i guess i should do a bigger scan |
| 22:43 | <Hixie> | and see what that tells us |
| 22:43 | <SadEagle> | Philip`: if it's not any effort to compute, what about 8859-5? |
| 22:44 | <annevk> | Hixie, yeah... :) |
| 22:44 | <annevk> | Philip`, and while you're at it, how about some stats? :) |
| 22:44 | <annevk> | (on charsets in general) |
| 22:45 | <Philip`> | SadEagle: That wouldn't be any effort, though it'll take a short while while I send grep through 3GB of HTML again :-) |
| 22:45 | <Hixie> | Philip`: you should set up a hadoop cluster for yourself :-) |
| 22:46 | <Philip`> | annevk: I suppose I could just collect all the HTTP content-type and meta content-types and summarise that |
| 22:47 | <Philip`> | Hixie: I only have about one machine so it wouldn't be a very good cluster, and it'd be fast enough by itself if it wasn't part of someone else's tiny cluster that's running an automated theorem prover several million times and using up all the CPU time :-) |
| 22:47 | <annevk> | nice nice |
| 22:50 | <gsnedders> | Hixie: having it clearer would always be nice |
| 22:50 | <Hixie> | gsnedders: please reply to anne saying what you would suggest to make it clearer then :-) thanks :-) |
| 22:51 | <Philip`> | SadEagle: I see about 7 that claim to be iso-8859-5 |
| 22:51 | <Hixie> | Philip`: hah |
| 22:52 | <Hixie> | i guess i should actually implement these character encoding algorithms and use those to work out the encodings, huh |
| 22:52 | <annevk> | to be honest, my initial tokenizer implemention did not make the distinction either, but that was quickly rectified when i did the parsing stuff |
| 22:52 | <Hixie> | i'll do that after lunch |
| 22:52 | <Hixie> | back later |
| 22:52 | <annevk> | i still don't like them being non-deterministic |
| 22:52 | annevk | -> the wire |