00:01
<zewt>
Philip`: asking for commentary on WebGL isn't quite the same as asking for it on typed arrays, either
01:49
<miketaylr>
abarth: do you know if anyone has tried to contact winktoolkit people?
01:49
<miketaylr>
i've worked with them in the past on some compat stuff
01:49
<abarth>
miketaylr: nope, i just saw the bug report today
01:50
<miketaylr>
k, i'll email the guy and at least point him to the discussion
03:40
<smaug____>
volkmar: ping
03:41
<smaug____>
volkmar: could you ask someone managing moznet to allow more connections from the hotel
03:44
<smaug____>
or gavin or ehsan
03:44
<ehsan>
I don't have control over moznet
03:45
<smaug____>
ehsan: could you perhaps ping someone
03:45
<ehsan>
smaug____: I only know gavin
03:45
<ehsan>
don't know anybody else with the privs :/
03:46
smaug____
files a bug to fix the situation
03:46
<ehsan>
smaug____: fwiw, I can't get on moznet myself :(
03:46
<smaug____>
oh
03:48
<volkmar>
I will ask on #developers
03:49
<gavin>
I can't do that myself
03:49
<gavin>
file a server ops bug
03:51
<smaug____>
volkmar: ehsan bug 739516
03:51
<ehsan>
smaug____: thanks
03:53
volkmar
is so glad to use his server to connect to IRC :)
04:15
<zcorpan>
annevk: don't you mean s/non-streaming/streaming/ ?
04:15
<zcorpan>
oh wait
04:18
<zcorpan>
annevk: didn't we agree on having the non-streaming api on string/arraybuffer?
04:53
<zewt>
zcorpan: no reason to do that
04:53
<zewt>
i mean, there's no benefit to having separate streaming and non-streaming apis
04:54
<zewt>
turning new TextEncoding('foo').encode(s) into s.encode('foo') is really not much of a win for having more API
04:55
<zewt>
s/Encoding/Encoder/
04:55
<zewt>
bed
05:21
<zcorpan>
Hixie: https://www.w3.org/Bugs/Public/show_bug.cgi?id=16524 is about recent canvas stuff
05:27
<Sirisian>
He might not be awake
05:28
<Sirisian>
If Hixie is then I forgot to ask him about this: https://www.w3.org/Bugs/Public/show_bug.cgi?id=11517
05:56
<zcorpan>
so apparently there's no spec anywhere saying how percentage heights work inside tables
05:56
<zcorpan>
yay!
08:12
<MikeSmith>
http://lists.w3.org/Archives/Public/public-iri/2012Mar/0070.html
08:12
<MikeSmith>
[[
08:12
<MikeSmith>
RFC format: There is ongoing developing some tooling and workflow which will allow us
08:12
<MikeSmith>
to generate versions of our internet drafts and RFCs which have HTML and
08:12
<MikeSmith>
PDF alternatives.
08:12
<MikeSmith>
]]
08:13
<MikeSmith>
to generate HTML versions
08:14
<MikeSmith>
I wonder how that's different from the generated HTML that tools.ietf.org already has
08:15
<MikeSmith>
and I wonder in which decade it might actually happen that the sources are maintained in HTML and the plain-text generated from that
08:15
<MikeSmith>
readability WTF
08:16
<annevk>
still no update on whether Unicode is okay huh?
08:16
<annevk>
IETF is hilarious
08:16
<MikeSmith>
there's still a few years left in this decade
08:16
<MikeSmith>
this is like the meeting of the Ents man
08:17
<MikeSmith>
I don't mean this particular message
08:18
<MikeSmith>
large irony here about multiple nit-picking editorial bugs filed against HTML spec and other W3C specs about "readability" issues
08:20
<MikeSmith>
as Jesus said, "wilt thou say to thy brother, Let me pull out the mote out of thine eye, and, behold, a beam is in thine own eye"
08:47
<Velmont>
Hmmm. Having all the other tests reside inside one test that only does setup seems to be a quite nice way to do tests.
08:48
<Velmont>
Looking at http://w3c-test.org/webapps/WebStorage/tests/submissions/Infraware/storage_local_key.html -- seems a nice way to do it.
08:53
<MikeSmith>
http://tools.ietf.org/html/draft-montenegro-httpbis-speed-mobility-00
08:54
<MikeSmith>
see the parts where is says "Same SPDY" and the parts where is doesn't
08:54
<MikeSmith>
e.g., "It would maintain the integrity of the layered architecture."
08:55
<MikeSmith>
implication being that SPDY does not "maintain the integrity of the layered architecture"
08:57
<MikeSmith>
also, see the one about "Protect our precious bodily fluids."
08:58
<annevk>
layered architecture sounds painful
08:59
<MikeSmith>
wow, it took 6 people to write this draft
08:59
<hsivonen>
just what everyone needed: SDPY getting the Web Socket treament at the IETF
09:00
<annevk>
MikeSmith: to write an introduction to the topic you mean
09:00
<annevk>
MikeSmith: section 2 is non-existent, and 3 and 4 are boilerplate :)
09:01
<annevk>
nice tweet hsivonen
09:01
<MikeSmith>
it's not a spec... it's not a requirements document... it's...
09:02
<MikeSmith>
mini-manifesto?
09:03
<MikeSmith>
is this mostly intended as implicit criticism of SPDY or what?
09:03
<zcorpan>
Velmont: i think you're supposed to use setup(func) and ideally let the script produce the same amount of tests even if the setup fails
09:04
<Velmont>
zcorpan: Yes, that's what I've been doing a few times before. But it's not always easy to make small unit tests fail like that.
09:04
<Velmont>
zcorpan: Quite often the error condition is also "timed out" which is quite slow.
09:05
<hsivonen>
any word from Apple or Opera about SPDY?
09:06
<zcorpan>
Velmont: i don't see how a wrapper test helps with timed out :-)
09:07
<Velmont>
zcorpan: Well, -- the first test will most probably fail, -- but the rest 20 tests will then never actually get created, -- so it will help not having all of those time out.
09:07
<annevk>
hsivonen: it's on our roadmap, though I suspect we might delay implementing if the IETF gets a go at it because we'd just lose a lot of time
09:07
<hsivonen>
gotta love it when a user with Firefox issues has 15 add-ons
09:07
<Velmont>
zcorpan: Well, -- not really a big problemo, -- but I'm always looking at different ways of writing the tests.
09:07
<hsivonen>
I guess I'm going to install those add-ons in a VM...
09:08
<MikeSmith>
e.g., does "Any HTTP 2.0 proposal should leverage the reliable transport and not attempt to replicate functions generally accepted as addressed by other layers." there because they see SPDY as replicating functions of other layers?
09:08
<hsivonen>
annevk: yay IETF for delays
09:08
<hsivonen>
MikeSmith: could have something to do with SPDY blending the TLS and HTTP layers
09:08
<annevk>
i was like "hurray" when I saw the HTTP 2.0 stuff come by
09:08
<zcorpan>
Velmont: not creating the other tests probably confuses spartan. i'm not sure but i think the framework doesn't wait for tests to time out if the setup fails
09:09
<MikeSmith>
hsivonen: OK
09:09
<annevk>
then made a little dance for being only tangentially involved in network standards
09:10
<hsivonen>
also, part of why SPDY works so well is that the TCP is a leaky abstraction, so multiplexing in SPDY is more efficient that letting IP multiplex multiple TCP connections
09:10
<MikeSmith>
I see
09:11
<hsivonen>
but AFAICT, SPDY does use TCP for "reliable transport"
09:34
<zcorpan>
iirc timj said he had tried to argue in Kronos that exposing endianness differences wouldn't work on the Web, but he was alone in the group to think so
09:36
<annevk>
:(
09:37
<annevk>
I have the impression he's often the sole sane voice there
09:37
<annevk>
I wonder how many years of experience the other people in that group have with browsers
09:47
<Velmont>
Ms2ger: Hey, -- for testharness.js; I'm translating old NETWORK_ERR etc from the testcases to NetworkError. And then check if the implementation threw a newstyle DOMException (by checking if it has a type), if so: check that type == NetworkError, and code = 27. -- If old-style, only check for the code.
09:48
<Velmont>
Ms2ger: Is the approach fine? https://github.com/jgraham/testharness.js/pull/1
09:48
<Velmont>
hrmf
10:00
<hsivonen>
aargh. what's the deal with bank sites having bad HTML
10:00
<hsivonen>
86 blank lines before the doctype. looking good.
10:01
<hsivonen>
very long meta keywords. check
10:01
<hsivonen>
coverage: global
10:01
<hsivonen>
money well allocated to SEO instead of reasonable HTML
10:02
<hsivonen>
ancient Netscape 4.x window resize fix
10:02
<hsivonen>
an empty script
10:12
<hsivonen>
cool. Hixie has actually explained what I need to do to fix this bank: http://lists.whatwg.org/htdig.cgi/whatwg-whatwg.org/2012-February/034869.html
11:02
<annevk>
two more decoders to go
11:03
<annevk>
actually, make that 3
11:03
<annevk>
big5, big5-hkscs, and gbk
11:06
zcorpan
wonders if it's reasonable to not support the unitless length quirk in calc()
11:08
<zcorpan>
seems mozilla already does that
11:09
<zcorpan>
webkit also
11:13
<zcorpan>
which makes sense, since calc can use numbers e.g. 2 * 5px
11:28
<zcorpan>
fixed
11:46
MikeSmith
wonders what encoding the #entmoot channel uses
11:46
<zcorpan>
hmm, my fix was wrong
11:51
<Philip`>
MikeSmith: I thought Entish was a non-written language (since if it was written then it could be read too hastily, and also trees don't have hands), so it's probably something like a very-low-frequency Vorbis encoding
11:51
<MikeSmith>
heh
11:51
<MikeSmith>
that explains the humming
11:54
<Philip`>
Oh, Wikipedia says I'm wrong - they adopted Quenya later, so presumably it'd be written in standard tengwar
11:59
<MikeSmith>
Philip`: so there is yet hope
12:04
<MikeSmith>
can somebody look at http://platform.html5.org/ in Chrome for me and tell me if rendering of the list in the Legend section is borked?
12:16
<Ms2ger>
MikeSmith, looks good, but the green flag seems a little wider than the rest
12:18
<MikeSmith>
Ms2ger: thanks
12:34
<annevk>
I think I can probably describe all encodings using this index concept
12:34
<annevk>
except perhaps for gb18030 as it is silly
12:51
<MikeSmith>
annevk: kindly please change the date on the diffs doc to March 29
12:56
<annevk>
so things are going done before the end of the month?
12:56
<MikeSmith>
yeah
12:57
<annevk>
how about we change html5-diff after the rest is done?
12:57
<annevk>
ah nm
12:57
<annevk>
i'll just do it now
12:58
<annevk>
MikeSmith: done
12:58
<MikeSmith>
arigatou
13:08
<annevk>
IANA is so crappy
13:08
<annevk>
http://www.iana.org/assignments/charset-reg/Big5-HKSCS has dead links
13:09
<annevk>
oh yeah, I should blame the group for maintaining that I guess, not "IANA"
13:09
<annevk>
otherwise someone might have a fit over the difference
13:28
<annevk>
hmm forgot hkscs is layered on top of big5
13:28
<annevk>
so big5-related stuff is pretty much a mess
13:34
<hsivonen>
time flies. the xtranormal Web Socket video is from December 2010
13:36
<annevk>
test
13:37
<annevk>
hmm so my connection is flaky, but not that flaky
13:42
<Ms2ger>
Velmont, your patch is on my todo list... I'll try to get to it tonight
13:42
<Ms2ger>
Sorry for the dealy
13:42
<Velmont>
:-)
13:42
<Ms2ger>
delay, even
13:47
<remysharp>
should I be using a <progress> element to indicate file upload, even though I know the total size before the file is sent (this is the bit that's confusing me in the spec)?
13:52
<annevk>
remysharp: yes
13:52
<annevk>
remysharp: that would be a determinate progress
13:53
<remysharp>
annevk: would it be fair to say that meter could go up and down, whereas progress can only go up?
13:53
<remysharp>
so I could (as the spec suggests) use meter to track disk space available
13:55
<annevk>
disk measurement is not some task that the user waits for
13:55
<annevk>
progress is related to some kind of activity, meter is just measuring things
13:56
<remysharp>
so, that's kind of a yes to my up/down analogy :) -- just makes it a little easier for me to remember
13:56
<remysharp>
I think I just saw an example of meter being used to measure activity in a google group
13:56
<remysharp>
which sounds right by my understanding now
13:58
<annevk>
yeah, I might temporal activity
13:58
<annevk>
meant damnit
13:59
<remysharp>
I liked my up/down thing - temporal reminds me of x-files for some reason :) (as requested, sense left at the door)
13:59
<annevk>
yeah dunno if up/down works in all cases, but it might
14:00
<remysharp>
I quietly suspect you're right
14:03
<annevk>
yay Orange
14:03
<annevk>
(WHATWG list has context)
14:49
<annevk>
hmm
14:50
<annevk>
both Opera and Chrome use surrogates for big5-hkscs
14:50
<annevk>
Gecko does not
15:29
<annevk>
you'd think that iterating over 20000 code points is not that hard
15:29
<annevk>
I guess it's not in benchmarks and that's why everything is slow
15:34
<annevk>
in IE they do indeed appear identical mappings
15:35
<annevk>
I guess the next step is some intersection research
15:37
<gsnedders>
annevk: Iterating over 20k codepoints is fairly quick. Doing stuff with them isn't.
15:39
<annevk>
good point
16:17
<dglazkov>
good morning, Whatwg!
17:57
<hsivonen>
is there a reference for Hixie's pre-IETF Web Sockets not having the problem with transparent proxies?
18:05
<Hixie>
what's "the problem with transparent proxies"?
18:05
<Hixie>
(there could be several)
18:11
<hsivonen>
Hixie: the proxy cache poisoning thing
18:11
<hsivonen>
Hixie: the one that was the big security thing at the IETF
18:11
<Hixie>
i'm not aware of a cache poisoning issue
18:12
<Hixie>
the thing that was a "big security thing at the IETF" as far as i can tell was a different thing for each person who thought there was a thing
18:12
<Hixie>
so that doesn't narrow it down for me :-)
18:13
<hsivonen>
Hixie: http://www.ietf.org/mail-archive/web/hybi/current/msg04744.html
18:14
<Hixie>
oh that's not necessarily cache poisoning. That's just the cross-protocol attack.
18:14
<Hixie>
my version of the protocol wasn't vulnerable to that because my protocol required sending what looked like two requests, the first one well-formed, and the second one noise
18:15
<Hixie>
and only if the two requests were treated as a single request by the proxy, and sent to the server together, and then the entire response sent back together as well, would the client send any more bytes
18:17
<Hixie>
(i never really understood why people were so suddenly concerned by that attack; by the time adam put out that paper, the issue had long been resolved in the spec)
18:30
<hsivonen>
oh. Chrome has migrated about:foo to chrome://foo vanity URLs
18:30
<hsivonen>
like Opera
18:30
<hsivonen>
except chrome: means something totally different in Gecko-based browsers
18:43
<jwalden>
that's probably better, actually; about:* has weird security properties, would be better if the privileged stuff that lived under there in Gecko were instead under a completely different protocol that were flat-out off-limits
18:44
<jwalden>
whether about:* could ever be made only the safe stuff in Gecko, tho, dunno
19:43
<Hixie>
TabAtkins: is 'image-rendering' gone?
19:44
<Hixie>
TabAtkins: (re https://www.w3.org/Bugs/Public/show_bug.cgi?id=12044#c2)
19:53
<annevk>
hsivonen: Hixie: I think the reason people were concerned is because nobody shipped or implemented the latest version of Hixie's spec, which was published just before the fork happened
20:05
<Hixie>
at least safari did
20:05
<Hixie>
but seems like implementing what i wrote would have been better than disabling it ;-)
20:05
<Hixie>
anyway that's history
20:05
<Hixie>
lunch!
20:09
<annevk>
no disagreement there
20:14
<karlcow>
http://inkdroid.org/journal/2012/03/27/cc0-and-git-for-data/
20:18
<hober>
Any comments on or suggestions for http://www.w3.org/html/wg/wiki/User:Eoconnor/ISSUE-200 before I send it in?
20:24
<annevk>
if we are going to revisit where <legend> can be used, we should use it for <details> and <figure> again...
20:25
<annevk>
"In fact, the author conformance restriction on <legend> *helps* authors to avoid butting heads with <fieldset>'s (unchangeable) weirdness." I think we can still make this better by describing the <fieldset> behavior in terms of some new CSS concept
20:25
<annevk>
but it looks more like UI design these days has no need for <fieldset>
20:29
<hober>
I'll add a note suggesting that those interested could propose 'some new CSS concept' to www-style; thanks
20:35
<TabAtkins>
Hixie: Punted to level 4.
20:39
<TabAtkins>
Are we generally against the init*Event functions?
20:40
<smaug____>
TabAtkins: yes
20:40
<smaug____>
in general
20:40
<smaug____>
not sure if there are some special cases
20:40
<smaug____>
(I think there aren't)
20:41
<TabAtkins>
What's the replacement?
20:42
<kennyluck>
constructors?
20:42
<TabAtkins>
That's what I thought.
21:21
<jgraham>
Yeah init*Event shouldn't be used in new features
21:22
<TabAtkins>
kk. The question came up regarding initTransitionEvent
21:30
<jgraham>
TabAtkins: In unrelated news you now have me wondering about the biochemistry of the female reproductive system
21:30
<TabAtkins>
...
21:30
<TabAtkins>
How?
21:30
<jgraham>
http://twitter.com/#!/tabatkins/status/184523386486128641
21:31
<TabAtkins>
Ah, yes.
21:31
<TabAtkins>
The female reproductive system is slightly acidic.
21:31
<jgraham>
Ah
21:31
<TabAtkins>
Thus at least part of the taste.
21:31
<jgraham>
Oh, that's quite famous isn't it
21:31
<jgraham>
The acidity, not the taste
21:31
<jgraham>
Although probably that too
21:32
<TabAtkins>
I dunno if acid snatches are famous.
21:34
<jgraham>
I was under the vauge impression -- and this is very much not my specialist area -- that people who tried to closely monitor their reproductive cycle for one reason or the other took pH into account
21:35
<TabAtkins>
Oh, right. Yes.
21:35
<TabAtkins>
That does vary across the cycle.
21:50
<jgraham>
Help, I have been on the internet too long. I see http://browserquest.mozilla.org/ and think of http://mozillaquest.com/
21:51
<jgraham>
(I assume one isn't supposed to think of that?!)
22:08
<Hixie>
http://www.w3.org/html/wg/wiki/ChangeProposals/notitle_captions#Details
22:08
<Hixie>
really?
22:08
<Hixie>
we're actually going to put an example in the spec _encouraging_ nested figures?
22:15
<nesta_>
Hixie hi?
22:15
<nesta_>
http://www.w3.org/html/wg/wiki/ChangeProposals/notitle_captions#Details
22:16
<nesta_>
cool! :)
22:16
<nesta_>
thnx!
22:43
<zewt>
ugh, File API added "oneTimeOnly" when nobody has actually defined that in any even vaguely meaningful way?
22:43
<zewt>
very poor
23:24
<Hixie>
woot, i finished canvas v5
23:24
<Hixie>
tomorrow i'll try to write replies to all the relevant e-mails and bugs
23:25
<TabAtkins>
Btw, thank you for the clearClip()!
23:26
<Hixie>
resetClip()?
23:26
<Hixie>
it was on the list of requests :-)
23:26
<TabAtkins>
Yeah.
23:26
<Hixie>
i added pretty much everything people asked for more than once and which didn't rely on infrastructre that didn't exist already
23:27
<Hixie>
(along with some infrastructure for the latter)
23:27
<Hixie>
(so that we can add them next time)
23:27
<Hixie>
TabAtkins: i'm gone the rest of the week as far as lunches go, and haven't been able to catch hober, but i'm free all next week so if you manage to catch him and convince him to come for lunch feel free to calendar me
23:28
<Hixie>
bbl
23:29
<TabAtkins>
kk