00:51
<annevk>
jgraham, I discussed this a week ago or so
00:51
<annevk>
jgraham, Philip` suggested I patch his processor script
00:51
<annevk>
btw, the W3C multipage version has better chosen splits than the WHATWG one
00:56
<othermaciej>
annevk: the page boundaries are different?
01:02
<annevk>
yeah
01:02
<annevk>
see e.g. forms
01:18
<othermaciej>
odd
01:20
<karlcow>
http://blog.threepress.org/2010/04/02/designing-ebooks-for-epub-reading-engines/
01:29
<Kronuz>
hello
01:31
<Kronuz>
hey, I'm trying to make a html satinize function using html5lib satinizer in python, I'm trying this code: http://pastebin.org/133583
01:32
<Kronuz>
but then I tried to input this: sanitizeHtml('<div style="text-align:center;margin-top:20px;"></div>')
01:33
<Kronuz>
and I get: '<div style="margin-top: 20px;"></div>'
01:33
<Kronuz>
where did my text-align go? and why? :(
01:35
<Kronuz>
...also, is there a way I can tell html5lib to add / to the self closing elements, such as printing <img ... /> or <br /> if I input <img ...> or <br> ?
01:36
<Hixie>
why would you want to do the latter?
01:36
<Hixie>
(not sure why you lose text-align, maybe html5lib is doing CSS sanitising also and strips text-align?)
01:37
<Kronuz>
Hixie, yep, it is I suppose, but why is it stripping it?
01:37
<Hixie>
not sure
01:37
<Hixie>
jgraham, annevk: yt?
01:37
<Kronuz>
and about the <... /> it's 'cause I want to have valid xhtml
01:37
<Hixie>
html5lib outputs html, not xhtml
01:37
<Hixie>
i think you have to pick an xml mode output if you want valid xhtml output
01:38
<Kronuz>
I would have guessed text-align was a "default" css property
01:38
<Hixie>
(ending <img> with /> is the least of your problems if you want valid xml output)
01:38
<Kronuz>
yep, well, I need a valid chunk, not the whole thing
01:39
<Kronuz>
I mean I wouldn't want it to output the doctype and stuff
01:39
<Hixie>
why are you using xhtml, anyway?
01:39
<Kronuz>
I'd be inserting it myself
01:39
<Hixie>
there's a lot more to outputting valid xhtml than just the doctype, namespaces, and />
01:39
<Hixie>
there's all kinds of crazy complications
01:40
<Kronuz>
anyway, the main problem right now is to understand what's wrong with my satinizer that doesn't let text-align (and who knows what else) thru ... :P
01:41
<Hixie>
yeah dunno about that
01:41
Hixie
isn't an html5lib guy
02:05
<Kronuz>
is there any way to make html5lib sanitize the css inside <style> tags?
02:06
<Kronuz>
or is there any way I can sanitize those later on using the html5lib css sanitizer?
03:10
<shepazu>
Hixie: what's this about views?
04:59
<Hixie>
i love how leif cites html4 then claims it says the exact opposite of what it says
05:03
<Hixie>
(i'd argue the case on the mailing list, but what html4 says seems somewhat academic)
08:22
gsnedders
waits for his hands to thaw so that he can type
08:22
<gsnedders>
It's a useful thing to be able to do
09:38
<Hixie>
hm
09:38
<Hixie>
is there any reason we shouldn't make <form> transparent and allow it in phrasing content elements?
09:38
<Hixie>
oh i guess <form> closes <p>, huh
09:42
<othermaciej>
so after reading the names for the algorithms that are alternatives to adoption agency, I am really curious what they do
09:42
<othermaciej>
and which one is which browser
09:42
<othermaciej>
I am guessing "incest algorithm" is IE
09:42
<Hixie>
yes
09:43
<othermaciej>
I don't know what Heisenberg or secret affair are
09:43
<Hixie>
Heisenberg is mozilla's old algorithm, which depended on packet boundaries
09:43
<othermaciej>
but I assume one is Firefox and the other is Opera
09:43
<Hixie>
and "secret affair" is opera, because it stores additional information beyond what you can see in the dom
09:44
<othermaciej>
I can see how adoption agency turned out to be the lesser evil
09:44
<Hixie>
e.g. <span style="a"> aaaaa <b style="b"> bbbbb </span> bbbb </b> will result in a DOM like <span> aaaaa <b> bbbbb </b></span> bbbb
09:44
<Hixie>
but it'll use the styles from <b> when rendering the last text node
09:46
<othermaciej>
that reminds me vaguely of how WebKit handles blocks inside inlines
09:46
<othermaciej>
(though the evil magic there is solely in the render tree)
09:46
<othermaciej>
but that could be just because I am currently trying to fix a bug with inline continuations
09:47
<othermaciej>
specifically how they interact with the accessibility tree
09:47
<othermaciej>
I may have to give up and ask hyatt for help
10:30
<Hixie>
huh, neither mozilla, nor safari, nor google reader complained about my xml error in my test atom document
10:30
<Hixie>
it took opera to point it out to me
10:30
<Hixie>
good work atom implementors!
10:32
<annevk>
am now Hixie
10:32
<Hixie>
annevk: some guy was asking html5lib questions
10:32
<annevk>
oh, I wouldn't have known either; it's been too long since I played with that
10:33
<othermaciej>
Hixie: I'm pretty sure Atom gets parsed in a tag soupey way
10:33
<othermaciej>
in Safari
10:33
<othermaciej>
our RSS/Atom parsing is kind of bunk
10:55
<Hixie>
man the state of atom implementations is terrible
10:59
<Hixie>
ok bed time
10:59
<Hixie>
nn
11:08
<annevk>
is this issue-96 proposal thread worth a read?
11:08
<annevk>
it seems to start out with shelley not understanding why ARIA is not the final solution...
11:09
<othermaciej>
it was a pretty one-sided thread (Shelley vs. everyone else) but I'd say it's worth reading and perhaps you can add something useful
12:34
<gsnedders>
Hixie: not well-formed in what way?
18:21
<jgraham>
Kronuz left I guess
18:22
<jgraham>
Oh well for future referne you can make html5lib give trailing slashes but if you really want xml output you should use an xml serializer in some xml library
18:23
<jgraham>
And the CSS sanitizer is some horrible regexp that I want to delete
18:23
<jgraham>
(but can't really delete until it can be replaced by a proper CSS parser)
18:24
<jgraham>
But there is a whitelist of properties. I can't see why text-align wouldn't be on it though
20:33
<annevk>
the amount of things people try to infer from vague lines in old specs is amazing
20:33
<annevk>
pro tip: specs have bugs
20:41
<Dashiva>
Isn't that the point of vague specs, letting people interpret them freely? :)
20:56
AryehGregor
really wishes <input type=email> would catch on so sites actually accept e-mail addresses with + in them
20:56
<AryehGregor>
It's extremely sad how many sites reject such addresses.
20:56
<Dashiva>
No kidding
20:57
<Dashiva>
I've been told it's a "security risk"
20:59
karlcow
wonders which kind of security risk. Spam is another security risk ;)
20:59
<Dashiva>
Maybe they don't escape it properly, so the + becomes a space or something
21:00
<Philip`>
Dashiva: I've seen that bug in one site
21:00
<Philip`>
(An Intel one, I think)
21:00
<Philip`>
where it constructed a URL like "register.cgi?user=foo+bar⊙bc"
21:04
<Dashiva>
Especially find is the examples like sun
21:04
<Dashiva>
Where they also reject + in the email address in the form to send feedback about registration
21:05
<Dashiva>
(I'm not sure what I meant instead of 'find')
21:06
<Philip`>
Maybe you meant "fun"?
21:06
<Dashiva>
I guess
21:07
<Philip`>
Or perhaps you meant something like "finickity"
21:08
<Philip`>
which admittedly wouldn't have made sense