| 00:37 | <gsnedders> | annevk: the problem is implementing sucky standards. |
| 08:44 | <annevk> | gsnedders, implementing them is the first step |
| 09:15 | <Hixie> | woot, managed a whole three paragraph e-mail with equal line lengths throughout |
| 09:17 | <annevk> | in 512 characters? |
| 09:31 | Hixie | investigates <span ===> |
| 09:31 | <Hixie> | webkit: ignores any attributes starting with = |
| 09:32 | <Hixie> | opera does too |
| 09:33 | <Hixie> | wow, so does firefox |
| 09:34 | <annevk> | ie7 doesn't |
| 09:34 | <annevk> | <span ===? gives SPAN ="==" |
| 09:36 | <Hixie> | yeah that's what the spec will say |
| 09:36 | <Hixie> | except it'll now be a parse error too |
| 09:38 | <Hixie> | actually no |
| 09:39 | <Hixie> | the spec will require <span ===> to become a SPAN with an '=' attribute with value '='. |
| 09:39 | <annevk> | ie specialcases = in the name state |
| 09:39 | <annevk> | to directly go to the attribute value state |
| 09:39 | <Hixie> | effectively |
| 09:40 | <annevk> | it's actually weirder |
| 09:41 | <annevk> | <span =x"="x> versus <span =x"="x"> |
| 09:42 | annevk | can't really make sense out of it |
| 09:43 | <Hixie> | <span a=b =x"="x> for me is being treated as a text node |
| 09:44 | <Hixie> | wow yeah that's some confusing shit right there |
| 09:44 | <Hixie> | glad we don't care enough about that edge case |
| 10:16 | <Hixie> | <x a=& b="& "> has no parse errors, but <x c="&"> does |
| 10:16 | <Hixie> | that seems bad |
| 10:17 | <Hixie> | and fixing it would be a bitch unless we just allow &" and &' anywhere. |
| 10:20 | <annevk> | that doesn't seem too harmful |
| 10:24 | <Hixie> | hm, found an easy solution |
| 10:24 | <Hixie> | which will probably be a pain to implement, but oh well |
| 10:24 | <annevk> | rly? |
| 10:24 | <annevk> | hmm |
| 11:20 | <Hixie> | how do we feel about adding entities? |
| 11:21 | <annevk> | i think it makes sense to add the ones IE already has |
| 11:22 | <annevk> | and MathML at some point... |
| 11:23 | annevk | sitll considers named entities rather useful |
| 11:43 | <Philip`> | Is it intentional that <foo bar=> is not a parse error? |
| 11:43 | <Hixie> | yes |
| 11:44 | <gsnedders> | I too would like to see MathML entities, FWIW |
| 11:44 | <Hixie> | see the comment i just added about entities |
| 11:45 | <gsnedders> | ah |
| 12:20 | <Philip`> | Hmph, every time I update my tokenisers to match the spec, the spec changes |
| 12:23 | <Hixie> | well right now i'm responding to tokeniser and parser feedback |
| 12:23 | <Hixie> | so yes |
| 12:23 | <Hixie> | :-) |
| 12:23 | <Hixie> | bed time now though |
| 12:44 | <gsnedders> | Anyone have any ideas of how to parse a dict URI with a turing machine (i.e., breaking the spec) |
| 12:46 | Philip` | discovers that "<foo bar=''" now has more parse errors than before |
| 12:48 | <Philip`> | Python html5lib now fails 21 tests, so I guess I shouldn't commit the updated tests |
| 12:49 | <gsnedders> | it's awfully easy to define something that's impossible to parse with ABNF |
| 13:15 | <Philip`> | Hixie: "After attribute value (quoted) state" -> "Reconsume the character in thebefore attribute name state." is missing a space |
| 18:36 | <annevk> | Is http://forums.whatwg.org/viewtopic.php?t=164 basically a request for CURIES? ... |
| 19:14 | <Lachy> | I like how he claims that the processing requirements for the new feature are "The same as the current <BASE> element " |