00:12
<Philip`>
rubys: Since you said you were interested a few days ago: http://canvex.lazyilluminati.com/svn/tokeniser/ has my current, uh, meta-tokeniser(?) code
00:13
<Philip`>
(It can run the tokeniser in OCaml, and can create one in C++)
00:13
<Philip`>
((Both are just missing non-numeric entity support since that wasn't very interesting and I haven't added it yet))
00:16
<rubys>
((you like to talk parenthetically, don't you?))
00:17
<rubys>
I've downloaded it and run it.
00:17
<Philip`>
(It's a bad habit of mine :-( )
00:19
<othermaciej>
too much lisp coding?
00:19
<rubys>
make_cpp looks fairly small; by implication make_py or make_rb would be too. Of course, an equivalent to tokeniser.cpp would also be necessary.
00:22
<Philip`>
The entity handling seems to require the most work in the language-specific code - it would be nice if that could be done more generically, like the rest of the state machine, but I've not really looked into that
00:53
Philip`
tries to work out what modifications to the generated C++ code would help efficiency easily
01:11
<Philip`>
http://canvex.lazyilluminati.com/misc/statestats.txt
01:11
<Philip`>
There's nearly as much double-quoted attribute value as there is plain text
01:15
<Philip`>
(Single-quoted is twice as common as unquoted, and double-quoted is twenty-five times more common than single-quoted)
01:15
<Philip`>
I guess the people who like XML syntax should be happy that unquoted values are so uncommon
01:19
<Philip`>
Assuming nobody has multiple doctypes, I found 1621 in 2522 pages, which is fortunately about the same as the 41%-with-no-doctype that Hixie reported
02:12
<othermaciej>
that's uncommon?
07:50
<zcorpan_>
why does HTMLCollection.namedItem() check for .name on some elements as opposed to any html element?
07:52
<zcorpan_>
is it that we want HTMLDocument.commands only look at id and not name?
07:53
<zcorpan_>
i'd rather it was more consistent with the other collection attributes... hmm
07:56
<zcorpan_>
oh. nevermind. e.g. <table>.rows doesn't look at name attributes
09:03
<zcorpan_>
is [[Get]] e.g. this?: forms[0]
11:29
<Dashiva>
zcorpan_: That's the most common use, at least
11:31
<zcorpan_>
Dashiva: ok
11:36
<Dashiva>
(on a lower level it's an internal property access, overridden to look for something other than the exact property; the details aren't too relevant, though)
14:35
<zcorpan_>
should window.frames look for svg foreignObjects?
14:40
<zcorpan_>
i.e., do foreignObjects create a nested browsing context?
14:47
<zcorpan_>
i.e. external foreign objects
15:03
<hsivonen>
losts and lots of continue and break in the parser...
15:03
<hsivonen>
just a step away from goto programming...