00:01
<Hixie>
why not use <a rel=prefetch>?
00:02
<TabAtkins>
So you don't have to answer questions like "What does <a rel=prefetch media=audio> do?".
00:02
<Hixie>
that's getting the priority of constituencies backwards
00:02
<TabAtkins>
Only insofar as you believe you can come up with an answer that'll make sense for users.
00:03
<Hixie>
i don't see why <link> would change the answer's sense-making ability
00:04
<TabAtkins>
We already know what effect @media has on <link rel=stylesheet>, so it carries over automatically.
00:04
<Hixie>
i'm pretty sure no UA implements that but sure
00:05
<Hixie>
(for rel=prefetch i mean)
00:05
<TabAtkins>
Sure.
00:06
<Hixie>
kinda tempted to dump the whole prefetch feature
00:27
<mcarter>
hello
00:29
<mcarter>
quick question about the websocket spec: given code new WebSocket("ws://example.com?") should the user agent include the trailing "?" in the request url?
00:39
<MikeSmith>
jgraham: hey, I guess you're the one who wrote the original version of https://developer.mozilla.org/en/Introduction_to_using_XPath_in_JavaScript
00:39
<MikeSmith>
if so, thanks
00:40
<MikeSmith>
it's pretty good
00:44
<Hixie>
mcarter: yes
00:47
<mcarter>
Hixie, the spec says "If /url/ has a <query> component, then append a single U+003F QUESTION MARK character (?) to /resource name/, followed by the value of the <query> component.", but "ws://example.com?" has no query component
00:53
<Hixie>
mcarter: i believe it actually does but that basically depends on the definition of the parser which is rather up in the air currently
00:54
<mcarter>
Hixie, hmm, well, the more important question I have then is if the UA should reject the ws connection if the sec-websocket-location header doesn't contain the trailing '?'
00:54
<Hixie>
yes
01:14
<mcarter>
Hixie, I've had some issues with existing cgi-derived technology + websocket (like python's wsgi for instance) which doesn't actually give you the full URL, rather it gives you things like QUERY_STRING which is an empty string in either case (trailing '?' or not)
01:14
<mcarter>
Hixie, obviously these things can be rewritten; they should just give expose the full request url. Its too bad they were designed like that..
01:16
<Hixie>
yeah
01:16
<Hixie>
in websockets the path and query are all just part of the resource name
01:16
<Hixie>
so that should be less of an issue
02:38
<mcarter>
Hixie, I'm mostly just interested (when its easy) in making websocket servers work, even when they depend on existing HTTP stacks (like eventlet/websocket which uses python wsgi http://bitbucket.org/which_linden/eventlet/src/tip/eventlet/websocket.py )
03:13
<Hixie>
mcarter: simplest solution is to never have a websocket resource with a trailing ?, then the problem never occurs
03:14
<mcarter>
Hixie, oh sure, but that just exposes the constraint to the end-user
03:15
<mcarter>
Hixie, I'm not particularly worried.
03:15
<mcarter>
Hixie, Its just one of those bugs I can imagine popping up once every blue moon, causing someone spending days, confused, trying to fix it
03:21
<Hixie>
yeah
03:21
<Hixie>
just check for it and raise a warning when you spot it, to make it easy to fix
03:24
<mcarter>
Hixie, sound advice
03:39
<mcarter>
Hixie, well, thats kind of the problem... dunno how to check for it on the server side.
03:39
<mcarter>
short of rewriting the http stack I mean
05:50
<annevk5>
svn.whatwg.org is slow
06:52
<annevk5>
I wonder where I should dump my work on XHR
06:52
<annevk5>
I guess I'll make some more tests first
07:11
<annevk5>
ooh, <meta http-equiv=link> would also mean HTMLMetaElement supporting LinkStyle
07:12
<annevk5>
clearly we should just drop the feature and avoid needless complexity
07:24
<annevk5>
I think I found another bug in XMLHttpRequest
07:24
<annevk5>
http://dev.w3.org/2006/webapi/XMLHttpRequest/#document-response-entity-body bullet point 2 should say "or" rather than "and", no?
07:24
<annevk5>
it's either null, text/xml, application/xml or ends in +xml
07:25
<aho>
yes
07:25
<aho>
should be or
07:25
<annevk5>
meh
07:26
<annevk5>
i guess after a while nobody reads the document anymore
08:08
annevk5
tries his luck
08:08
<annevk5>
will emailing the WHATWG list go faster than filing bugs?
08:08
<annevk5>
we'll find out shortly
08:15
<annevk5>
okay, not shortly
08:38
<nessy>
is there a list admin of whatwg around?
08:39
<nessy>
no worries - I will re-edit the email and re-send
08:42
<annevk5>
that's prolly the easiest
10:04
<annevk5>
making tests for the conditional get requirement was somewhat harder than anticipated
17:26
<erlehmann>
oh, i just found this http://lastweekinhtml5.blogspot.com/
17:26
<erlehmann>
who writes it? is there any history behind it or just random trolling?
17:27
<gsnedders>
erlehmann: That's the great mystery about it, who writes it. Just random trolling, though.
17:28
<erlehmann>
it's not exactly my kind of humour, though i frequent an imageboard that shall not be named
17:29
<erlehmann>
but there is a class of dumb jokes, where you need specific knowledge to GET it.
17:30
<erlehmann>
it reminds me of the picture of ahmadineschad pointing at the camera with the caption NO U²³⁸
17:30
<gsnedders>
It's a lot less funny than it used to be
17:30
<Workshiva>
It was somewhat clever in the beginning, then it devolved into random insults and profanity
17:31
<Workshiva>
And near the end it almost seemed to include some not-anti-whatwg sentiment, oddly enough
17:36
<erlehmann>
that technique of finding the worst possible photographs of people makes me chuckle
18:05
<erlehmann>
haha, remember that imageboard, whose admin told me that <video> was not ready yet? people are now using GIF to upload porn: http://krautchan.net/board/b/thread/1955067
18:05
<erlehmann>
facepalm
18:06
<erlehmann>
and each "video" uploaded this way is like 3 or 4 mb. for seconds … oh, the humanity.
18:09
<Philip`>
erlehmann: Not even APNG?
18:09
<Philip`>
Surely it'd be better with full 24-bit colour
18:09
<erlehmann>
342x257 GIFs
18:10
<erlehmann>
;_;
18:12
<Workshiva>
erlehmann: oldnews.tk
18:13
<erlehmann>
-_,-
18:40
<Workshiva>
"Tickets also can have a target milestone, which would be either 4-week or 6-month."
18:40
<Workshiva>
I know I'm a week late, but is hybi really planning another 6 months on websocket?
19:08
<jgraham>
Workshiva: The idea (kind of) is that things that are hard to change are fixed in a few weeks in the hope that they are included in {browser}.next releases
19:09
<jgraham>
In six months more complex things can be added
19:09
<jgraham>
As long as they don't affect back-compat
19:09
<jgraham>
Not everyone agrees on all the details though
19:09
<Philip`>
Also it means you don't have to say "no" to people's pet features, you just have to say "six months"
19:33
<Workshiva>
Seems to be a lot "some people were talking at the f2f and decided x, but we aren't going to tell anyone why"
20:26
<Philip`>
Workshiva: It's about consensus, not about whys
22:52
<Steve^>
If it used time it will be 2010-07-06 for 6th July and not ambiguous