| 11:40 | <annevk> | http://www.phpied.com/wp-content/uploads/2008/02/canvas-pie.html is nice |
| 11:42 | <Hixie> | hard to argue against the .style in that |
| 11:46 | <annevk> | heh, i wasn't even thinking along those lines |
| 11:47 | <annevk> | just that it was a nice example of <canvas> usage that degrades pretty good to UAs that don't support <canvas> or non-visual UAs |
| 15:47 | Philip` | doesn't like how the parser spec uses "... and ...", "..., and ...", "..., and then ...", "..., then ..." for exactly equivalent sentences in different places |
| 15:48 | <Philip`> | So far I've got /Act as if an? (start|end) tag (?:token )?with the tag name "(\S+)" (?:and no attributes )?had been seen(?:, and then|, and| and|, then) reprocess the current token\./ to handle all the variations I've seen |
| 15:48 | <gsnedders> | Philip`: I don't think that's in the en-gb-x-Hixie spec, better complain to the editor |
| 15:49 | <Philip`> | I guess the language was designed for humans, which makes it harder for me :-( |
| 15:49 | <zcorpan_> | Philip`: since you're inhuman? |
| 15:50 | <annevk> | heh, you're actually trying to generate code from the English text? :) |
| 15:50 | <Philip`> | zcorpan_: No, but I'm trying to create an inhuman :-) |
| 15:52 | <Philip`> | annevk: It seems to be working alright so far - mostly it's a big table of English->code, but there are lots of repeated and nearly-repeated phrases so it saves a lot of code writing |
| 15:52 | <Philip`> | and I can be fairly surely it precisely matches the spec, including when the spec changes in the future, which is good since I'm lazy and don't want to have to check it carefully :-) |
| 15:53 | <annevk> | i guess you should comment on the inconsistencies |
| 15:54 | <Philip`> | I could add a code->English translation function, to rewrite that section of the spec with no inconsistencies :-) |
| 16:00 | <zcorpan_> | we should define how Hixie English should be parsed |
| 16:00 | <gsnedders> | is http://ian.hixie.ch/bible/english not the place for that? |
| 16:01 | <zcorpan_> | yes but it's very incomplete |
| 16:01 | <gsnedders> | en-gb-x-hixie 5? |
| 16:02 | Philip` | wonders if it's sane to convert all the "generic R?CDATA parsing algorithm" bits into separate new insertion modes, so he can get rid of the bits where it needs to wait for the next token |
| 16:02 | <annevk> | heh |
| 16:02 | <Philip`> | ...and the skip-next-token-if-it's-a-newline bit too |
| 16:02 | <annevk> | the /TR/ version actually uses <html lang=en-US-x-Hixie> |
| 16:02 | Philip` | hopes there aren't any other such places |
| 16:02 | <gsnedders> | annevk: US!? |
| 16:02 | <annevk> | and the WHATWG version has en-GB-Hixie |
| 16:03 | <gsnedders> | so the /TR version is in an unspecified language. yuk. |
| 16:05 | <annevk> | en-US-x-Hixie is more correct than en-GB-Hixie |
| 16:05 | <gsnedders> | annevk: "more correct" makes no sense. |
| 16:05 | <gsnedders> | correctness is a binary value, you are either correct or you are not |
| 16:05 | <gsnedders> | annevk: but why? |
| 16:05 | <Philip`> | The English words "abort", "about", "above", "accepting", ... "youngest", "your" are to be interpreted as described in the Oxford English Dictionary [OED]. |
| 16:06 | <Ketsuban> | You forgot "sausage". |
| 16:06 | Ketsuban | promptly apologises for the Blackadder reference. |
| 16:08 | <zcorpan_> | gsnedders: hixie english is both en-GB-x-Hixie and en-US-x-Hixie |
| 16:08 | <zcorpan_> | gsnedders: en-GB-hixie and en-US-hixie are the old tags |
| 16:08 | <gsnedders> | zcorpan_: but there's no spec for en-us-[x-]hixie! :P |
| 16:09 | <gsnedders> | hmm, en-us-hixie is mentioned once in it |
| 16:09 | <zcorpan_> | point that out to Hixie :P |
| 16:09 | <gsnedders> | and not even with an -x :P |
| 16:10 | gsnedders | notes you need an en-gb-x-hixie parser to parse the en-gb-x-hixie spec |
| 16:11 | <Philip`> | You need an HTML parser to parse the HTML spec too |
| 16:11 | <gsnedders> | yeah, I know that. |
| 16:12 | <Ketsuban> | gsnedders: a standard en-GB parser is sufficient. |
| 16:12 | <gsnedders> | Ketsuban: most people have some variant of en-gb, though |
| 16:14 | <Ketsuban> | True. Language codes don't take into account idiolects (excepting of course en-GB-x-Hixie). |
| 16:56 | <Philip`> | Hooray, now my parser does <title> and implies <html> and <head> and <body> etc, and I think it might even do that correctly |
| 16:57 | <annevk> | we should change handling of <title> and </head> in HTML5 though |
| 16:57 | <Philip`> | If that can be done by copying-and-pasting text from other sections, then I'll be happy since I won't have to do anything |
| 16:59 | <annevk> | i guess it comes down to </head> being ignored but only being a parse error sometimes and <title> being treated like a normal element except for its content model |
| 17:22 | Philip` | wishes he could use the tree construction tests without first having to work out how to glue his tokeniser and tree-constructor together |
| 17:24 | <Philip`> | ...and fortunately I can, since I can just use some other tokeniser to convert the tests so they give the token stream instead of the source text |
| 17:25 | <Philip`> | ...but unfortunately I don't have a JSON parser in OCaml |
| 18:00 | <kig> | there is a json parser for ocaml, json-static or whatever it was |
| 18:01 | <kig> | (or at least there's one in our work repo) |
| 18:18 | <Philip`> | kig: Ah, thanks - json-static looks a bit scary with camlp4, but json-wheel looks like it ought to work without pain |
| 20:22 | <gsnedders> | <http://www.w3.org/TR/2008/PER-xml-20080205/>; — PER of Fifth Edition of XML 1.0 |
| 20:24 | <Philip`> | XML 1.0 5? |
| 20:25 | <AwayEagle> | That's purely editorial, I imagine? |
| 20:25 | <gsnedders> | AwayEagle: no, it isn't |
| 20:25 | <Philip`> | No |
| 20:25 | <Philip`> | http://recycledknowledge.blogspot.com/2008/02/justice-at-last-part-two.html |
| 20:27 | <gsnedders> | so _that's_ what that change is all about |
| 20:33 | <AwayEagle> | interesting. Doesn't XML NS specify QName, etc., grammar in terms of unicode categories anyway? |
| 20:33 | <gsnedders> | AwayEagle: depends whether you look at NS for XML 1.0 or NS for XML 1.1 |
| 20:34 | <AwayEagle> | gsnedders: sounds like I looked at the wrong one then :-) |
| 20:34 | <gsnedders> | AwayEagle: NS for XML 1.0 will have to be revised, FWIW |
| 20:35 | <gsnedders> | actually, it won't |
| 20:35 | <gsnedders> | actually, yes it will. |
| 20:35 | <gsnedders> | it refers to stuff in appendix B in XML 1.0 |
| 23:48 | Philip` | starts passing a few tests |