07:58
<hsivonen>
hmm. the “multiple of two” thing is informative in the HD canvas spec text
07:59
<hsivonen>
has anyone successfully deployed canvas backing stores with multiples of CSS px other than 1 or 2?
08:03
<MikeSmith>
hsivonen: HD canvas spec text?
08:31
<hsivonen>
MikeSmith: http://www.whatwg.org/specs/web-apps/current-work/multipage/the-canvas-element.html#resolution
08:32
<MikeSmith>
hsivonen: thanks
08:32
MikeSmith
reads
08:32
<MikeSmith>
man when did this happen
08:32
<MikeSmith>
I been out of it for quite a while I gues
08:32
hsivonen
wonders if Hixie mean “power of” rather than “multiple of”
08:33
<hsivonen>
I’ve lost track of what is happening, too.
08:33
<hsivonen>
s/mean/meant/
08:36
<hsivonen>
Interesting http://browser.yandex.com/
08:37
<annevk>
hsivonen: sounds like a bug indeed
08:37
<annevk>
hsivonen: iPhone has 4:1
08:38
<hsivonen>
annevk: huh? is iPhone’s devicePixelRatio gone up to 4, too?
08:39
<hsivonen>
*has
08:39
<annevk>
hsivonen: dunno about that, but each CSS pixel is four device pixels
08:40
<hsivonen>
you mean like 2 by 2 device pixels?
08:40
<annevk>
well yeah
08:40
<MikeSmith>
hsivonen: about Yandex browser, https://twitter.com/chaals/status/252679467862986752
08:40
<MikeSmith>
"Take a webkit base, add some Opera source, spice it up with Yandex taste, and serve on a cloud :)"
08:41
<hsivonen>
MikeSmith: how much of a joke does the :) part signify?
08:41
<MikeSmith>
not sure
08:42
<hsivonen>
Adding some Opera source to a WebKit base seems *very* surprising
08:42
<MikeSmith>
but doesn't the word "cloud" now always imply a joke?
08:42
<doublec>
I winder if he meant "sauce"
08:42
<hsivonen>
MikeSmith: not if it’s an Opera Mini competitor
08:42
<MikeSmith>
doublec: chaals is from Australia, so not a native English speaker...
08:44
<annevk>
from a press release I read they're using Opera Turbo; so maybe they licensed that somehow?
08:56
<MikeSmith>
hsivonen: have you have time for what should be a quick validator patch review?
08:56
<MikeSmith>
for NPE that was reported today
08:56
<MikeSmith>
from input@pattern datatype code
09:00
<MikeSmith>
ohI just now notice the fingerprint icon that Hixie added to some notes in the spec
09:01
<MikeSmith>
anybody know which browser engines currently implemented the Media Fragments URI stuff for temporal fragments?
09:01
<MikeSmith>
I know Gecko does
09:01
<MikeSmith>
can't remember about others
09:02
<MikeSmith>
doublec: ?
09:02
<MikeSmith>
caniuse nor nobody else tracking status of it as far as I can tell
09:03
<MikeSmith>
hmm Gecko and WebKit I guess
09:03
<MikeSmith>
http://www.w3.org/2008/WebVideo/Fragments/WD-media-fragments-impl/
09:03
<MikeSmith>
wonder about IE10
10:32
<doublec>
MikeSmith: gecko and webkit is what I know of
10:32
<doublec>
MikeSmith: Opera had a demo, not sure if they finished it
10:39
<zcorpan>
hsivonen: isn't power of two also wrong? i don't see why 3x3 device pixels per css pixel is obviously bad
10:40
<zcorpan>
but maybe i'm missing something
10:40
<annevk>
3^2 == 3x3 though
10:42
<zcorpan>
....so it is
10:43
zcorpan
tries to think of something to blame
10:43
<jgraham>
Uh, I don't think that's what hsivonen meant
10:43
<jgraham>
Or at least if he did I didn't understand him either
10:44
<zcorpan>
the sentence under discussion is "Ideally, the number of device pixels per CSS pixel would be a multiple of two."
10:44
<jgraham>
3 isn't a multiple of two (for reasonable meanings of "multiple")
10:45
<zcorpan>
but you'd have 9 device pixels per css pixel, not 3
10:45
<jgraham>
Which is still neither a power nor a multiple of two
10:46
<zcorpan>
ah so my thinking was right anyway
10:47
<zcorpan>
it should be n^2 not 2^n
10:47
<jgraham>
I think that Hixie means linearly pixels_device = 2^N pixels_css
10:48
<jgraham>
Or maybe he means 2N
10:48
<zcorpan>
he clearly *wrote* 2N, what he meant i have no idea :-)
10:48
<jgraham>
Makes more sense
10:49
<jgraham>
I don't even understand why this is "ideal"
10:49
<jgraham>
So maybe I should just stop talking
11:19
<hsivonen>
MikeSmith: I can review a patch, sure.
11:19
<MikeSmith>
hsivonen: ok
11:19
<MikeSmith>
thanks
11:19
<MikeSmith>
it's only a couple lines, so I'll post it here I guess
11:19
<MikeSmith>
gist
11:20
<MikeSmith>
cause seems to be lack of null check in the Rhino code
11:20
<MikeSmith>
for which I filed a Mozilla bug
11:20
<MikeSmith>
but from validator side, the fix would seem to be this: https://gist.github.com/3809647a
11:21
<MikeSmith>
passing an actual Context object instead of null
11:21
<hsivonen>
404 says github
11:21
<MikeSmith>
oh
11:21
<hsivonen>
their 404 page is pretty cool
11:22
<MikeSmith>
hey yeah
11:22
<MikeSmith>
will make you dizzy though
11:22
<MikeSmith>
ok will try to re-post it
11:23
<MikeSmith>
hsivonen: OK, https://gist.github.com/3811066 should work I think
11:29
<hsivonen>
MikeSmith: “Each call to enterContext() must have a matching call to Context.exit().”
11:29
<MikeSmith>
aha
11:29
<MikeSmith>
OK
11:29
<MikeSmith>
will fix that
11:29
<hsivonen>
MikeSmith: in a finally block
11:29
<MikeSmith>
OK
11:29
<hsivonen>
to make sure it gets called
11:30
<MikeSmith>
hai
11:31
<MikeSmith>
only reason I'm using enterContext() to begin with is that their API docs say simple "new Context()" is deprecated
11:31
<MikeSmith>
anyway, will fix it an e-mail to you
11:31
<hsivonen>
it’s not a good sign that Rhino is hosted on SVN and Google doesn’t find the javadocs on www.mozilla.org
11:32
<hsivonen>
the API is quite bizarre
11:32
<hsivonen>
enterContext() returns a Context, but .exit() is not on that object but on the class
11:32
<hsivonen>
apparently the class keeps a ThreadLocal somewhere
11:33
hsivonen
is not sure how great an idea it is for this stuff to support multithreading
11:33
<hsivonen>
considering that JS doesn’t have facilities for shared memory threads
11:33
<hsivonen>
(at least not that I’m aware of)
11:35
<MikeSmith>
hsivonen: the whole API seems really baroque to me
11:38
<hsivonen>
I wonder if Yandex ships NPAPI Flash or Pepper Flash
11:44
<MikeSmith>
hsivonen: OK please lemme know if the following is sane
11:44
<MikeSmith>
https://gist.github.com/3811142
11:49
<hsivonen>
MikeSmith: yeah, but you don’t need two try blocks. You only need one try and you can have both catch and finally on it
11:49
<MikeSmith>
ah yeah
11:49
<MikeSmith>
OK
11:57
<MikeSmith>
hsivonen: OK, the following should do it, right?
11:57
<MikeSmith>
https://gist.github.com/3811200
12:05
<hsivonen>
MikeSmith: yeah, r+. thanks
12:08
<MikeSmith>
hsivonen: thanks much
12:57
<annevk>
MikeSmith: shouldn't it be ex.exit() ?
12:57
<annevk>
MikeSmith: sorry, cx.exit()
12:58
<annevk>
MikeSmith: or is it a static method?
12:58
<MikeSmith>
yeah
12:58
<MikeSmith>
so called on the class
12:58
<annevk>
weird
12:58
<MikeSmith>
mind you I don't really understand that code well
12:58
<MikeSmith>
that's just exactly what the API docs say to do
12:58
<MikeSmith>
annevk: there is lots more weirdness in that code
12:59
<annevk>
but they don't mean you invoke it on the instance?
12:59
<MikeSmith>
no, they don't
12:59
<MikeSmith>
not as far as I can tell
12:59
<annevk>
I mean I sometimes talk about XMLHttpRequest.responseText...
12:59
<annevk>
okay
13:00
<annevk>
in other news, writing new slides is boring
13:00
<MikeSmith>
yeah man
13:00
<annevk>
my trick was going to be to talk about 10 lines of HTML for an hour, but actually working that all out... meh
13:01
<hsivonen>
annevk: what’s the talk?
13:01
<MikeSmith>
boringness of it is why I wait to makes slides til 3am of the morning when I need to present
13:01
<annevk>
it's for http://fronteers.nl/congres/2012
13:01
<annevk>
my plan is to talk about complexity
13:02
<MikeSmith>
annevk: fwiw, my experience in seeing you present is that you can in fact actually take 10 lines of code and talk about them for an hour and actually make it interesting the whole time
13:03
<hsivonen>
annevk: I suggest talking about browsing contexts then
13:04
<annevk>
oh man, browsing contexts :)
13:04
<hsivonen>
In my experience, the best way to be productive is not even trying to ever fix anything browsing context-related
13:04
hsivonen
is right now trying to fix something browsing context-related
13:48
<Ms2ger>
irc.w3.org down?
13:49
<MikeSmith>
Ms2ger: yeah
13:50
<reggna>
Ms2ger: MIT is having some problems.
13:50
<MikeSmith>
dunno what's going on because I can''t even get on w3c IRC to harass the systems team
13:50
<MikeSmith>
nor e-mail
13:50
<MikeSmith>
some MIT-wide outage, I think
13:50
<reggna>
15:30:58 < w3c> Network problems at MIT are causing failures connecting to W3C servers hosted there. MIT admins are working on the issue.
13:51
gavinc
mutters something about being awake at 6:30 am on a monday for no reason
13:51
<Ms2ger>
Better than 6:30 AM on a Sunday
13:52
<gavinc>
Ms2ger: Very true
13:54
<darobin>
MIT is back
15:21
<volkmar>
darobin: is http://respec.specifiction.com/js/profiles/w3c-common.js the latest respec version?
15:29
<darobin>
volkmar: nope, but hang on, I'm on a call
15:33
<darobin>
volkmar: repo for the latest is at https://github.com/darobin/respec/
15:33
<darobin>
volkmar: linking to the latest is from http://www.w3.org/Tools/respec/respec-w3c-common
15:36
<volkmar>
darobin: great 'partial' is now supported
15:36
<darobin>
volkmar: yes, and a bunch of other stuff too
15:36
<darobin>
the code is 95% new, but it should be backwards compatible
15:37
<darobin>
that is, if you haven't been doing dirty hacks that monkey patch the internals :)
15:38
<volkmar>
and I can use sequence
15:38
<volkmar>
great
15:45
<darobin>
volkmar: if you find issues, please use the GH tracker
16:35
<dglazkov>
good morning, Whatwg!
18:23
<Hixie>
cabanier: please avoid cross-posting e-mails to the whatwg list and another list -- because the whatwg list is subscriber-posts-only, it causes the threads to fragment and confuse people only on the whatwg list
18:23
<Hixie>
cabanier: better to either do it all in the whatwg list or do it all in the other list and then send a summary or pointer to the whatwg list
18:35
<cabanier>
hixie: OK. I assumed Dirk had wanted to post to both
18:35
<Hixie>
cabanier: he shouldn't cross-post either :-) (but i think he actually posted to both separately)
18:36
<Hixie>
(or at least that's now i interpreted his PS, i didn't check)
21:11
<krit>
Hixie: Yes, posted it a second time after a request of Rik ;)
22:39
<Hixie>
anyone have an opinion on <input type=text trim=""> ? (trims leading and trailing space characters in the control's /value/)
22:39
<Hixie>
if so, put them in https://www.w3.org/Bugs/Public/show_bug.cgi?id=17972
22:40
<Hixie>
i'm likely to add it soonish
22:43
<Hixie>
anyone got IE handy? I need http://software.hixie.ch/utilities/js/live-dom-viewer/saved/1811's output
22:43
<Hixie>
and i'm too lazy to get out my laptop with the VMs...