| 00:03 | <jgraham__> | Philip`: At some point it would be nice to have the data behind the phase transition diagram for the tree construction (the .dot file or whatever) |
| 00:03 | <jgraham__> | But right now I'm going to sleep |
| 00:04 | <Philip`> | jgraham__: http://canvex.lazyilluminati.com/svn/tokeniser/ has a possibly out of date version of the code for generating the graph (and more) |
| 00:16 | <Philip`> | (graph_gen.ml in particular produces the .dot file, based on treeconstructot_spec.ml which is generated by treeconstructor.pl from section-tree-construction.html which is an XMLified version of the section from the HTML5 spec) |
| 00:16 | <Philip`> | s/t/r/ |
| 10:16 | Hixie | will use this channel to note URLs to tables |
| 10:16 | <Hixie> | i hope no-one minds |
| 10:16 | <Hixie> | http://broads-authority.gov.uk/boating/navigating/tide-tables.html |
| 10:17 | <Hixie> | http://www.usability.com.au/resources/tables.cfm |
| 10:25 | <Hixie> | http://www.gutenberg.org/files/17128/17128-h/17128-h.htm |
| 11:07 | <Hixie> | http://www.gutenberg.org/files/21341/21341-h/21341-h.htm#metrical_feet |
| 11:11 | <Hixie> | http://www.gutenberg.org/files/19598/19598-h/19598-h.htm |
| 11:11 | <Hixie> | http://broads-authority.gov.uk/boating/navigating/tide-tables.html |
| 11:11 | <Hixie> | http://www.socialsecurity.gov/policy/docs/statcomps/eedata_sc/2004/table01.html |
| 11:12 | <Hixie> | http://joeclark.org/dossiers/PDFUA/PDFUA-tables-1.html |
| 11:43 | <Lachy> | I'm trying to find a way to reprhase this requirement in selectors api: "If the user agent also supports some level of CSS, the implementation should support the same set of selectors in both these APIs and CSS." |
| 11:44 | <Lachy> | I want it to say that UAs should support the same selectors, but allow for exceptions in case some selectors are introduced that don't work for CSS, but do for selectors api, or vice versa |
| 11:45 | <Lachy> | any suggestions? |
| 11:59 | <zcorpan> | to me it sounds like something that should be non-normative... |
| 12:04 | <Lachy> | yeah, that's what I was thinking |
| 12:29 | <Lachy> | I think this is better "For user agents that support of CSS, it is recommended that they support the same set of selectors for use in both these APIs and CSS." |
| 12:32 | <hsivonen> | so IE8 uses a bugus mime type application/x-hatom in WebSlices discovery... |
| 12:48 | <zcorpan> | Lachy: isn't "RECOMMENDED" an rfc2119 keyword? |
| 12:49 | <Lachy> | yes, it is |
| 12:50 | <Lachy> | but it allows for the requirement to be ignored |
| 12:50 | <Lachy> | if there are valid reasons to do so |
| 12:50 | <zcorpan> | same as "SHOULD" |
| 12:50 | <Lachy> | yes |
| 12:50 | <zcorpan> | so your change is merely editorial :) |
| 12:51 | <Lachy> | well, I couldn't think of a non-normative way to phrase it that didn't use either "should" or "recommended" |
| 12:51 | <Lachy> | but the existing draft in SVN was actually a MUST level requriement |
| 12:51 | <Lachy> | so it's more than editorial |
| 12:51 | <zcorpan> | ah |
| 12:52 | <zcorpan> | "User agents are strongly encouraged to..." |
| 12:52 | <zcorpan> | if you want it non-normativ |
| 12:52 | <zcorpan> | e |
| 12:52 | <zcorpan> | (or s/strongly //) |
| 12:52 | <Lachy> | that might work, though what are the advantages or disadvantages of making it non-normative? |
| 12:53 | <zcorpan> | is it something you'd put in a testsuite? |
| 12:53 | <zcorpan> | (though "should"s are often not in testsuites anyway...) |
| 12:53 | <Lachy> | it would be difficult to test with a test suite |
| 12:54 | <Lachy> | though, not impossible |
| 12:54 | <zcorpan> | indeed |
| 12:54 | <zcorpan> | might be a useful thing to test |
| 12:54 | <zcorpan> | dunno |
| 12:54 | <zcorpan> | don't really have an opinion on the matter |
| 12:54 | <zcorpan> | just trying to help with wording :) |
| 15:13 | <hsivonen> | hmm. Pingback interop doesn't look so great |
| 15:13 | <hsivonen> | pingback to WordPress requires <a href to appear in lower case in the blog post at my end |
| 15:31 | <Philip`> | It looks like it probably requires lowercase a, but it doesn't care about the href |
| 15:32 | <Philip`> | i.e. <a title="http://.../"> and <a http://.../> would work just as well |
| 15:39 | <zcorpan> | perhaps we should make <a http://...> parse into <a href='http://..'> |
| 15:40 | <Philip`> | Why? |
| 15:40 | <zcorpan> | convenient? |
| 15:41 | <Philip`> | Doesn't seem that hard to write href="" :-p |
| 15:41 | <Philip`> | particularly compared to the cost of updating tutorials and tools |
| 15:42 | <Philip`> | I see one person writing <a http://www.addthis.com/bookmark.php?pub="thatgrrl&url="; expr:href="" &title=" + data:post.title" target="_blank" title="Bookmark using any bookmark manager!"> but that's about it |
| 15:43 | <Philip`> | Also <a mailto:foo=bar⊙gc> would be kind of ambiguous |
| 15:43 | <hsivonen> | this charset alias stuff is crazy but fascinating in a way |
| 15:44 | <Philip`> | Sounds like the web |
| 15:45 | <hsivonen> | here's an idea for a testing project: extracting all the charset aliases supported by Gecko, generating test pages for each and checking which aliases are actually supported by IE, Safari and Opera |
| 15:46 | <hsivonen> | the aliases for the old Apple encodings seem to be particularly inconsistent in different implementations |
| 15:47 | <hsivonen> | according to http://philip.html5.org/data/charsets.html almost no one uses them anyway... |
| 15:48 | <Philip`> | There's a million times more pages on the web than what I looked at, so "almost no one" multiplies into quite a lot of people |
| 15:48 | <hsivonen> | another research item: figuring out the requirements for the visual and logical iso-8859 Hebrew and Arabic variants |
| 15:49 | <hsivonen> | Philip`: well, according to your numbers, <$mtpublishcharset$> is more popular than x-mac-thai :-) |
| 15:50 | <Philip`> | According to my numbers, the difference is not significant given the sample size, so you can't tell which is more popular :-) |
| 15:54 | <hsivonen> | implementations don't agree whether the chaset name for MacUkranian should contain the word ukraine or ukranian, whether there should be the x- prefix and whether there should be a hyphen after mac |
| 15:55 | <Philip`> | (Also my sample is quite strongly biased towards certain languages that are not Thai) |
| 15:55 | <Philip`> | (Er, my sample isn't, but my population is) |
| 15:56 | <Philip`> | (compared to the entire web) |
| 15:56 | <Philip`> | (I'm probably messing up the statistical terms anyway) |
| 15:57 | <hsivonen> | my local build of Validator.nu parser now knows about 1194 charset names (when certain ones have been banned) |
| 15:57 | <hsivonen> | that's just crazy when utf-8 should be the one name to rule them all |
| 15:59 | <Philip`> | Until n years in the future when a new character encoding that solves all the problems with UTF-8 will become popular and should rule over all others? |
| 16:00 | <hsivonen> | problems, what problems? :-) |
| 16:00 | <hsivonen> | (utf-8-nfc :-) |
| 16:01 | <Philip`> | There needs to be a URI-based namespacing system for private extensions, otherwise people will end up conflicting and it'll all go horribly wrong |
| 16:02 | <Philip`> | It's no good just allocating a flat block of numbers and telling people to do what they want in that range |
| 16:28 | <hsivonen> | fixing all these charset details is a mess |
| 17:21 | <Philip`> | hsivonen: http://hsivonen.iki.fi/rdf/ - s/Martial/Martian/ |
| 17:22 | <hsivonen> | Philip`: thanks fixed |
| 17:29 | <krijnh> | hsivonen: also, s/on error<A/on error <A |
| 17:29 | <krijnh> | (last paragraph) |
| 17:45 | <hsivonen> | krijnh: thanks fixed. |
| 17:53 | <Philip`> | Looks like half the SVG files on Wikipedia come from Inkscape |
| 17:54 | <Philip`> | and 25% from Adobe Illustrator |
| 17:55 | <Philip`> | Also one of them says <!-- lol jews did wtc --> |
| 17:55 | <Philip`> | I guess vandalism isn't restricted to content pages :-( |
| 17:56 | <Philip`> | (Mostel_HUAC.svg, in case anyone cares) |
| 17:57 | <takkaria> | heh |
| 18:01 | <KevinMarks> | someone described MS as having 'red state' developers |
| 18:03 | <KevinMarks> | ie lots of in-house intranet devs, who are their primary customers, and who have built sites predicated on everyone having the same MS browser, as that's what IT installs |
| 18:11 | <Philip`> | http://philip.html5.org/data/svg-xmlns.txt |
| 18:11 | <Philip`> | There's a lot of xmlns:svg (from Inkscape, I think) but no <svg:...> at all |
| 18:18 | <Philip`> | Hmm, Opera renders http://upload.wikimedia.org/wikipedia/en/2/23/Yonkers_seal.svg |
| 18:18 | <Philip`> | and Firefox doesn't |
| 18:18 | <gsnedders> | Saf doesn't |
| 18:19 | <Philip`> | Opera 9.5 doesn't |
| 18:19 | <Philip`> | (9.2 does) |
| 18:19 | <Philip`> | I guess that's just a bug |
| 18:20 | <hsivonen> | namespaces are tough |
| 18:21 | <Philip`> | and it affects about 2-3% of the SVG images on Wikipedia |
| 19:22 | <andersca> | Hixie: ping! |
| 20:08 | <eseidel> | Philip`: the SVG is wrong :) |
| 20:08 | <eseidel> | Philip`: it needs a namespace |
| 20:08 | <eseidel> | glad to see 9.5 has become more strict |
| 20:08 | <Hixie> | andersca: here |
| 20:10 | <andersca> | Hixie: hi! is this a typo? |
| 20:10 | <andersca> | A browsing context can be associated with an application cache. A child browsing context is always associated with the same browsing context as its parent browsing context, if any. |
| 20:10 | <andersca> | should it be "A child browsing context is always associated with the same application cache..." ? |
| 20:11 | <Hixie> | er yes |
| 20:11 | <andersca> | that makes things more clear :) |
| 20:11 | <Hixie> | please do e-mail the list about that one :-) |
| 20:11 | <andersca> | will do |
| 20:12 | <Hixie> | thanks |
| 20:13 | <andersca> | sent |
| 20:50 | <hsivonen> | w3.org is now ahead of whatwg.org for html 5 on google |
| 20:50 | <Hixie> | indeed |
| 20:50 | <Hixie> | unsurprising |
| 20:59 | <hsivonen> | Hixie: if confidence is Certain and a meta is seen for a different encoding (after alias resolution), shouldn't that be a parse error? |
| 20:59 | <hsivonen> | I though it already was but now I don't see it in the tree builder spec |
| 21:09 | <Hixie> | hsivonen: it's an error anyway, since you can't have more than one encoding declaration |
| 21:11 | <hsivonen> | Hixie: that doesn't cover the case where it became confident due to HTTP or BOM and meta disagrees |
| 21:11 | <Hixie> | i thought i'd made that non-conforming too |
| 21:11 | <Hixie> | yes, it is |
| 21:11 | <Hixie> | 3.7.5.4. Specifying the document's character encoding |
| 21:11 | <Hixie> | first bullet point |
| 21:13 | <zcorpan_> | Hixie: in the live dom viewer you can get ie8 working by using try-catch with the ie8 approach in the catch block |
| 21:14 | <Hixie> | there's a separate ie8.html file in that directory at the moment |
| 21:14 | <hsivonen> | Hixie: ok. the natural way to implement that is as parse error |
| 21:15 | <Hixie> | hsivonen: that makes sense |
| 21:15 | <zcorpan_> | yes, but it would be more useful if it worked in the normal version |
| 21:15 | <zcorpan_> | giving 2 duplicate links every time doesn't scale |
| 21:15 | <zcorpan_> | s/2// |
| 21:15 | <Hixie> | well, get them to fix ie8 :-) |
| 21:15 | <Hixie> | i don't want to fix the main one to work around bugs in unrelased products |
| 21:16 | <zcorpan_> | ok |
| 21:17 | <hsivonen> | I'm starting to suspect that we are going to need a charset registry 5 so that so that each implementor doesn't need to figure out the magic alias mess |
| 21:18 | <hsivonen> | can someone suggest a good behavior-based heuristic for testing if a character decoder is ebcdic-based? |
| 21:18 | <Hixie> | what mess? |
| 21:20 | <hsivonen> | Hixie: mapping TIS-620 to Windows-874. mapping GB_2312-80 to GBK. KS_C_5601-1987 to windows-949-2000. x-x-big5 to big5, etc. |
| 21:20 | <Hixie> | some of those will presumably end up in the spec, but i'm hoping that most are not widespread enough to require support |
| 21:20 | <hsivonen> | also, if the Gecko alias file and the ICU aliases are any indication of reality, the IANA registry of aliases is not the whole story |
| 21:41 | <Hixie> | http://www.flickr.com/photos/joeclark/192878174/ |
| 21:41 | <Hixie> | ...and other such tables |
| 21:41 | <Hixie> | seem to use indenting as a way of having "subrows" |
| 21:42 | <Hixie> | i wonder if we should support that explicitly somehow |
| 21:44 | <Hixie> | http://www.flickr.com/photos/joeclark/185786265/ is an insane table |
| 21:46 | <hsivonen> | how is x-user-defined supposed to work? |
| 21:48 | <hsivonen> | WebKit seems to implement x-user-defined as mapping to the first 256 code points of the PUA |
| 22:08 | <Hixie> | this indenting thing is really common |
| 22:09 | <andersca> | Hixie: URLs specified in the manifest attribute are relative to the document, right? |
| 22:10 | <hsivonen> | I sucks considerably that Java doesn't have a rewindable InputStream in the standard library |
| 22:10 | <Hixie> | andersca: yes, in html |
| 22:11 | <hsivonen> | one that doesn't require telling it how much to buffer ahead of time, that is |
| 22:11 | <hsivonen> | fortunately, I'm not the first person to need it, so I don't have to write it |
| 22:11 | <Hixie> | andersca: in xhtml, if there's an xml:base on the root <html> element, that is honoured |
| 22:11 | <andersca> | oh, interesting |
| 22:11 | <Hixie> | (assuming you support xml:base) |
| 22:11 | <Hixie> | ok my macbook pro is CLEARLY having overheating issues |
| 22:12 | <Hixie> | it keeps locking up when it gets even remotely warm |
| 22:12 | <hsivonen> | must be fun with manifest and <base> |
| 22:12 | <Hixie> | i wonder if the fans are busted or something |
| 22:13 | <andersca> | Hixie: how do you empty an application cache? is that something that the user agent decides when/how to do? |
| 22:15 | <othermaciej> | andersca: I think "Empty Cache" in the UI should do it |
| 22:15 | <andersca> | yeah |
| 22:16 | <Hixie> | empty cache does it |
| 22:16 | <Hixie> | there's also a way to trigger it |
| 22:16 | <Hixie> | but i forget what it is |
| 22:18 | <andersca> | swapCache doesn't clear it, or does it? |
| 22:22 | <Hixie> | ok so the way to clear a cache is to simply empty the manifest, but leave it on the server |
| 22:23 | <Hixie> | (and to remove the manifest attributes) |
| 22:24 | <Hixie> | bbiab |
| 22:24 | <andersca> | right |
