00:14
<Hixie>
anyone have an opinion on whether pushState() should take an absolute URI which the UA must then verify, or whether we should require that it be frefixed either with ? or with # and that it then replaces either the query part or the hash part?
00:17
<othermaciej>
Hixie: I think it should accept either relative or absolute and apply appropriate constraints to the resulting resolved URI
00:17
<othermaciej>
Hixie: a lot of "REST" style sites put the interesting info in path segments of the URI, not in a query or fragment ID
00:18
<Hixie>
we could allow the path part to change too, i guess
00:19
<Hixie>
so you're saying take the URI, resolve it relative to whatever it is you resolve URIs in scripts to, and then check the protocol/server/port are the same
00:20
<othermaciej>
yes, that would be my suggestion
00:20
<Hixie>
k
00:21
<Hixie>
sweet, my ilife '08 package arrived
00:21
<Hixie>
now go go el wire package!
00:34
kingryan
finds it interesting that http://www.whatwg.org/specs/web-apps/current-work/#outlines includes actual code
01:11
<Hixie>
hm
01:11
<Hixie>
an event to inform people that the fragment identifier has changed
01:12
<Hixie>
on...hashchanged?
01:17
<kingryan>
isn't there already an event for that?
01:17
<kingryan>
or maybe its just onscroll that gets emitted
01:21
<Hixie>
i don't think there's an event other than onscroll, no
01:21
<Hixie>
and onscroll doesn't always fire
01:25
<kingryan>
right
02:47
<deltab>
Hixie: would it be when navigation starts, or when the target fragment is shown?
03:22
<Hixie>
deltab: it fires when you're on a page and you do something to scroll the page to a new fragid
04:08
<mpt>
Hixie, how about when you load a page with a fragment in the first place?
04:09
<mpt>
ah, maybe that's already catered for
12:08
<annevk>
Hixie, http://annevankesteren.nl/2007/08/ajax-history people ask for replaceState()
12:09
<Philip`>
Hixie: About the parser performance data: I got significant speedups by measuring the expected lengths of strings (tag names, attributes names/values) and preallocating enough space for ~99% of those, so that's quite useful data to have too
12:10
<Philip`>
(I suppose data about attribute-value-length per attribute-name could be useful too - src and href and onclick etc are usually really long, but maybe you could easily preallocate space for all the shorter attributes)
12:13
<annevk>
Philip`, I mentioned this in a pm, but I suppose it can be discussed here as well, how about merging the <canvas> testsuites? It would be great if your testsuite became accessible to more people so they can contribute I think
12:18
<Philip`>
annevk: That sounds like a good thing to do
12:19
<annevk>
seems that you cover most, if not everything, of the tests I covered and more so using your stuff as base makes more sense now :)
12:19
<Philip`>
I can't think of any immediately obvious ways to merge all the existing ones, since mine is weird and puts everything in tiny boxes and tries to avoid ever having humans look at the results
12:20
<annevk>
it does mark the boxes either red or green which is enough for humans
12:25
<Philip`>
annevk: I think the problem is more with other tests like http://tc.labs.opera.com/html/canvas/getContext/2d/arc/004.htm and http://hixie.ch/tests/adhoc/html/canvas/005.html whose pass/fail outcome can't really be determined automatically
12:29
<annevk>
hmm, yeah, I suppose it makes sense that the testsuite consists of an automated part an a non-automated one
12:29
<Philip`>
(I think I've only got three which can't be done automatically in a perfect browser, plus two that can't be done automatically in Firefox/Opera because of bugs)
12:30
<Philip`>
((Bugs related to security of data: and drawImage and getImageData, in particular))
12:32
<Philip`>
(But IE/Safari can hardly do any of them automatically, since they lack getImageData or equivalent)
12:34
<annevk>
it makes some sense to require getImageData for the testsuite... it's a feature of <canvas> after all :)
15:10
<annevk>
someone suggested that hashchanged shouldn't be past tense to be more in line with other comments
20:01
<Hixie>
http://www.chewontech.com/2007/08/more-things-added-in-html-5.html is surprisingly well-written
20:15
<Philip`>
Hixie: Probably because they nicked it from http://www.ibm.com/developerworks/xml/library/x-html5/
20:16
<Hixie>
huh
20:16
<Hixie>
interesting
20:17
<Hixie>
the style of the chewontech page make it much easier to read than the ibm one
20:17
<Hixie>
to the point where i read the chewontech page but couldn't read the ibm one
23:13
<gsnedders>
how many times have I defended IE from people trying to make MS look bad?
23:14
<othermaciej>
is that a trick qustion?
23:14
<gsnedders>
othermaciej: a rhetorical one
23:14
<gsnedders>
complaining filter goes against the option in CSS2.1 of using hyphens or underscore this time, despite it predating CSS2.1
23:15
<othermaciej>
I think the CSS vendor prefix convention also predates CSS 2.1
23:15
<gsnedders>
othermaciej: I just looked it up. Couldn't find it in CSS2
23:15
<othermaciej>
although I don't know if it predates the IE filter property
23:16
<gsnedders>
IE filter was introduced in IE4
23:16
<gsnedders>
(which also predates CSS2)
23:16
<othermaciej>
the worse thing about filter is that SVG introduced a filter CSS property with an incompatible syntax
23:17
<othermaciej>
but I blame SVG for that, not Microsoft
23:19
<gsnedders>
it's often on a forum which I go to less and less, as I've been called an idiot so many times for believing that XHTML isn't the future, and that MS isn't totally shit
23:19
<othermaciej>
well, IE is really not that great in terms of standards complliance and has not advanced much in a long time
23:20
<othermaciej>
so I'm not sure all criticism of MS is off-base
23:21
<gsnedders>
a lot of it is
23:24
<Hixie>
there are a lot of people who criticise IE unfairly and incorrectly
23:24
<Hixie>
but there's a lot of criticism that IE could get fairly and accurately that it doesn't
23:24
<Hixie>
it's possible the two balance out :-)
23:24
<gsnedders>
also a lot who criticise the devs, cwilso especially
23:24
<Hixie>
chris is a fine man
23:24
<Hixie>
not the most active of chairmen, but that's another story
23:25
<Hixie>
someone give me a random number
23:25
<gsnedders>
42
23:25
<Hixie>
dom-events it is
23:25
<gsnedders>
(not really random, but hey)
23:26
<gsnedders>
in the case of some people in the WG they'll complain if he isn't active as chair, and if he is, they'll complain he isn't improving Trident
23:26
<Hixie>
he could be more active than he is without really eating into his Trident productivity much
23:26
<gsnedders>
I know, but they'll find reasons to complain at him someway or another
23:27
<gsnedders>
I remember one reaction I got around a year ago when I told someone Tantek was lead of IE5/Mac. Then my saying how old IE5/Mac was. They had assumed it was recent, and it was terrible because it was MS.
23:28
<gsnedders>
Hixie: why did you want a random number, though?
23:29
<Hixie>
i'm picking things to work on in the spec
23:29
<gsnedders>
then how do numbers map to things in the spec?
23:29
<Hixie>
42 was the 42nd folder on my IMAP server
23:29
<gsnedders>
ah.
23:29
<Hixie>
INBOX.input-for-whatwg-dom-events
23:30
<hendry>
:)
23:30
<Hixie>
ok
23:30
<Hixie>
<img alt>
23:30
<Hixie>
it seems there are three kinds of images
23:30
<gsnedders>
I wonder if Douglas Adams' ever knew what the consequences of trying to think of what a normal number would be
23:30
<Hixie>
images that are really just fancy alternatives to actual text
23:31
<Hixie>
images that are purely decorative
23:31
<Hixie>
and images that are intrinsic to the content but have no textual alternative
23:32
hendry
would agree with that
23:32
<gsnedders>
what about graphs?
23:32
<hendry>
gsnedders: vector graphics?
23:32
<gsnedders>
they have vital to the content, but can have a textual alternative
23:32
<gsnedders>
hendry: no, just images that contain graphs
23:32
<gsnedders>
anyhow, I must go sleep
23:32
<Hixie>
examples being a simple diagram or graph that can be described textually, a snapshot from a movie in a movie review or a diagram or graph that just repeats what the prose says, and a photo on flickr respectively