00:16
<zewt>
oh, jesus christ
00:16
<zewt>
apparently webkit allows navigating to blob URLs, and people are using it for things
00:17
<zewt>
https://github.com/eligrey/FileSaver.js/blob/master/FileSaver.js this idiot is navigating to a blob url to basically polyfill FileSaver
00:20
<zewt>
not entirely sure how since this code is pretty unreadable
00:22
<zewt>
or maybe it's trying to do that on firefox and doing something else in webkit
00:22
<zewt>
stripping down the code to figure it out
00:25
<zewt>
whew: looks like it's trying to do that in firefox, but is already broken (maybe firefox used to allow navigating to blob URLs and they stopped that)
00:25
<zewt>
<- terrified of anything that could make blob URLs more complicated and poorly specced than they already are
01:24
<sicking>
ng
01:25
<zewt>
wt
09:09
jgraham
wonders how to answer "would it be possible to add tests" without sounding rude
09:09
<jgraham>
The answer is something like "yes, of course. But someone needs to write them"
09:10
<jgraham>
zcorpan: Do you have any plans to add workers typed array tests?
09:11
<zcorpan>
jgraham: yeah i can do that
09:12
<jgraham>
zcorpan: Did you see the email on public-webapps-testsuite? There is a WebGL test but it is in some other format
09:12
<zcorpan>
i guess structured clone and transfer in general needs tests
09:12
<jgraham>
https://github.com/KhronosGroup/WebGL/blob/master/sdk/tests/conformance/typedarrays/typed-arrays-in-workers.html
09:12
<zcorpan>
yeah lemme reply
09:13
<jgraham>
OK, that's one easy solution :)
09:13
<jgraham>
I thought we had some tests for structured clone kicking around in some context, but I'm pretty sure they predate typed arrays
09:17
<zcorpan>
yeah
09:17
<zcorpan>
and probably predate transfer, too
09:20
<zcorpan>
reviewing the microsoft worker gives me the impression that they have tried to get as many test *files* as possible with as little effort as possible
09:20
<zcorpan>
which is a bit annoying
09:21
<zcorpan>
e.g. compare http://w3c-test.org/webapps/Workers/tests/submissions/Microsoft/Worker_ErrorEvent_filename.htm and http://w3c-test.org/webapps/Workers/tests/submissions/Microsoft/Worker_ErrorEvent_lineno.htm
09:21
<zcorpan>
(or maybe s/with as little effort as possible//)
09:22
<tobie>
wrote some basic tests for structured cloning a while back.
09:22
<tobie>
let me try to dig them out.
09:24
<jgraham>
To be fair the old Opera test thing with top.rr encouraged the same behaviour
09:25
<jgraham>
Or encouraged you to do things like passed = e.someProp == "expected1" && e.otherProp == "expected2" && [...]
09:26
<jgraham>
So it's hard to know if it's just a cultral thing where they are writing tests in the same way they always have
09:35
<smaug____>
zcorpan: then r- the tests ;)
09:35
<zcorpan>
smaug____: working on it
09:35
<zcorpan>
jgraham: yes
09:36
<zcorpan>
jgraham: i'm sure i have tests that do this sort of thing as well
09:37
<Ms2ger>
So, git wizards
09:37
<Ms2ger>
How do I make "git pull w3c" work?
09:37
<Ms2ger>
To pull from https://github.com/w3c/web-platform-tests
09:39
<jgraham>
Ms2ger: git remote add w3c git⊙gc:w3c/web-platform-tests.git
09:39
<jgraham>
If you don't already have a remote called w3c
09:40
<jgraham>
Otherwise git remote set-url w3c git⊙gc:w3c/web-platform-tests.git
09:40
<Ms2ger>
You asked to pull from the remote 'w3c', but did not specify
09:40
<Ms2ger>
a branch. Because this is not the default configured remote
09:40
<Ms2ger>
for your current branch, you must specify a branch on the command line.
09:40
<jgraham>
OK
09:40
<jgraham>
git branch --set-upstream {branchname} w3c/master
09:41
<jgraham>
(honestlky, I never use git pull. I always use git fetch and then explicity rebase)
09:43
<Ms2ger>
I have no idea what all of those are
09:43
<jgraham>
What all what are?
09:44
<jgraham>
git pull does two things by default: "git fetch" which pulls in all the commits from an upstream and "git merge" which tries to merge the current branch with the upstream branch it's tracking
09:44
<tobie>
zcorpan, jgraham: here: https://gist.github.com/tobie/3d81cd3d48859d2df1dd
09:45
<tobie>
LMK if that's of interest to you, in which case I'll submit them officially.
09:45
<Ms2ger>
I'm only pulling after my PR has been merged
09:45
<jgraham>
That is pretty silly behaviour because a) it is too easy to make a branch that's tracking the wrong thing and b) rebase is better than merge for unpushed changes
09:46
<Ms2ger>
And my unpushed changes are in a hg repo for my sanity
09:46
<jgraham>
So instead of doing the magic all-in-one operation, you can do each part independently (like in hg)
09:46
<jgraham>
git fetch, then git rebase {branch I think is the one this branch should be based on}
09:46
<annevk>
Hixie: I think I should give a response of sorts
09:47
<tobie>
Agreed. git fetch W3C and rebasing is a much more sane strategy.
09:47
Philip`
likes "git pull --rebase"
09:47
<jgraham>
You can also change the default behaviour of git pull to rebase
09:48
<zcorpan>
tobie: thanks
09:48
<Ms2ger>
So, I fetched
09:48
<Ms2ger>
Now I'm trying to push those commits to my fork
09:48
<jgraham>
(which is "git config branch.autosetuprebase always")
09:49
<jgraham>
git push {remote} HEAD:{remote_branch}
09:49
<jgraham>
Will push HEAD on your local checkout to {remote_branch} in the remote named {remote}
09:50
<jgraham>
(HEAD could be anything commit-like e.g. a branch name)
09:51
<tobie>
zcorpan: note I have zero recollection of why I implemented the structureClone helper function that way.
09:51
<jgraham>
(so you can do things like "git fetch w3c; git push ms2ger w3c/master:master" to update the master branch in the ms2ger remote with the latest commits from the w3c remote)
09:51
<Ms2ger>
*sigh*
09:51
<Ms2ger>
All I want to do is hg pull -u w3c && hg push
09:51
<Ms2ger>
Why is this so hard?
09:52
<jgraham>
It isn't :p
09:52
<Philip`>
Ms2ger: Because you make life hard for yourself by using two tools instead of one?
09:52
<Ms2ger>
Anyway, git pull w3c master and git push seems to have worked
09:53
<Philip`>
It only takes a few months to get used to how git does things
09:54
Ms2ger
tries to remember what he was going to do before git got in the way
09:55
<jgraham>
If all you are doing is shunting commits between different remote repos, I think you *really* don't want to use git pull
09:55
<jgraham>
But anyway, if you solved your problem, that's good :)
09:56
<Ms2ger>
Well, what did I want to do, then?
09:57
<Ms2ger>
Fetch did something, but apparently not enough
09:57
<SimonSapin>
eh, we found the packet of death that crashes our MTA … time to invoke http://xkcd.com/242/
09:58
<jgraham>
Ms2ger: fetch will bring all the remote commits into your local repository
09:58
<Philip`>
I think "git fetch" is equivalent to "hg pull" (without the "-u")
09:59
<Ms2ger>
Then how do I do hg update?
10:00
<SimonSapin>
Ms2ger: git merge origin/master (assuming the branch is master, and you want a merge rather than a rebase)
10:00
<jgraham>
I think the answer is "it depends exactly what you want to do"
10:00
<Ms2ger>
I don't want to merge
10:01
<SimonSapin>
although rebase and merge are the same if nothing have diverged. git calls that a "fast-forward merge"
10:01
<Ms2ger>
My tree is a subset of upstream
10:01
<jgraham>
If you want to checkout the upstream changes
10:01
<jgraham>
Then the safest option is git merge --ff-only
10:02
<SimonSapin>
Ms2ger: it’s still called a merge, although a "fast-forward" one
10:02
<jgraham>
Which will update your local branch but only if it is really a subset of upstream
10:02
<jgraham>
And throw an error if it isn't
10:02
<jgraham>
If you *do* have local commits
10:02
<Ms2ger>
A merge is when you have two heads, and you create a new commit with the two heads as parents
10:03
<SimonSapin>
Ms2ger: git calls that a non-fast-forward merge
10:03
<Ms2ger>
So git just calls things that aren't merges merges to confuse me?
10:03
<Ms2ger>
That sounds pretty much expected
10:03
<jgraham>
Well you are merging in the remote changes
10:03
<jgraham>
So it's not like it doesn't make sense
10:06
<jgraham>
(I don't tend to think in terms of "heads" with git. I just think in terms of branches)
10:06
<Ms2ger>
Not with my definition of merge :)
10:07
<Ms2ger>
Anyway
10:08
<Ms2ger>
Feel free to review :)
10:10
<jgraham>
I wonder why that didn't get a critic review
10:11
<Ms2ger>
It did
10:11
<Ms2ger>
This is an automatic message generated by the review at:
10:11
<Ms2ger>
http://critic.hoppipolla.co.uk/r/8
10:11
<jgraham>
Oh, I'm just not set up as a reviewer there :)
10:11
<Ms2ger>
tobie, and I'm pretty sure ES doesn't accept tests from people who aren't employed by a TC member
10:11
<zcorpan>
does https://github.com/w3c/web-platform-tests/pull/37 have a critic review?
10:13
<jgraham>
zcorpan: Does now: http://critic.hoppipolla.co.uk/r/9
10:13
<jgraham>
I forgot to transfer the filters to the new repo.
10:13
<tobie>
Ms2ger: that is just absurd.
10:13
<Ms2ger>
tobie, indeed
10:13
<jgraham>
tobie: Also, true
10:14
<tobie>
mmm.
10:14
<Ms2ger>
tobie, but no, I'm not going to bother fighting them
10:14
<tobie>
I might.
10:14
<jgraham>
Pretty sure the right thing to do is smuggle them in via jwalden or someone
10:14
<tobie>
I live 10 min away from Ecma HQ.
10:14
<Ms2ger>
I could do that
10:14
<Ms2ger>
But then I have to do work
10:14
<jgraham>
Well I mean the right thing to do is to take pitchforks and burning torches to ECMA
10:14
<Ms2ger>
Whereas it's much easier to just route around the damage
10:15
<Ms2ger>
jgraham, do the keyboard shortcuts actually work in Fx?
10:16
<jgraham>
Ms2ger: I think they might not. If they don't work in Chrome either I guess it will get fixed soon ;)
10:16
<Ms2ger>
Heh
10:16
<jgraham>
Otherwise, patches welcome (or I will look)
10:17
<tobie>
what license is test262 under?
10:17
<tobie>
can't find this info on the website.
10:18
<jgraham>
darobin, MikeSmith, tobie, zcorpan, Ms2ger, others: if you had filters on the html-tetssuite repo. in critic you will need to readd tehm for the web-platform-tests repo. Sorry about that.
10:18
<jgraham>
https://github.com/tronical/test262/blob/master/LICENSE
10:18
<jgraham>
Perhaps?
10:19
<tobie>
jgraham: ty.
10:20
<jgraham>
https://mail.mozilla.org/pipermail/test262-discuss/2012-February/000086.html is also relevant
10:20
<tobie>
OK, so I guess we should try to fix the contribution situation at Ecma (I'll give it a try), or fork the repo and pull it in somehow.
10:30
<zcorpan>
is there something that returns a File without user interaction?
10:30
<tobie>
Happy birthday darobin!
10:31
<tobie>
zcorpan: nope.
10:31
<zcorpan>
this is annoying
10:31
<tobie>
There's a thread about that being an issue for testing on the whatwg mailing list, afaik.
10:31
<darobin>
thanks tobie :)
10:31
<zcorpan>
ah
10:31
<tobie>
where I complain direly.
10:31
<jgraham>
Oh, happy birthday
10:32
<darobin>
thank you man :)
10:34
<zcorpan>
tobie: i can't find the thread. what should i search for?
10:34
<tobie>
zcorpan: webapps, actually: http://lists.w3.org/Archives/Public/public-webapps/2012OctDec/0675.html
10:34
<tobie>
and I must've dreamt about complaining.
10:35
<jgraham>
Maybe you are actually Victor Costan in disguise?
10:36
<tobie>
maybe.
10:36
<zcorpan>
tobie: thanks
10:38
<tobie>
jgraham: the more plausible answer is I failed at clicking send.
10:39
<tobie>
jgraham: or the content was so full of some of my signature garbled sentences it was trashed by a spambot.
10:40
<annevk>
darobin darobin darobin
10:40
<annevk>
happy b-day!
10:40
<Ms2ger>
I thought it was darobin's birthday just now...
10:43
<jgraham>
Maybe he's like the Queen
10:45
<tobie>
Ms2ger: Reached out to Ecma. we'll see how this goes. Will keep youposted.
10:46
<Ms2ger>
Ta
10:47
<tobie>
Do we have specs in the WebApps where we feel we have 100% test coverage?
10:47
<tobie>
Trying to validate our test coverage mode, here.
10:47
<tobie>
s/mode/model/
10:47
<Ms2ger>
Mm
10:47
<Ms2ger>
Progress events
10:48
<darobin>
annevk annevk! Thanks :)
10:51
<tobie>
Ms2ger: ty
10:51
<zcorpan>
if i'm reading webidl right, this should throw TypeError: postMessage(null, [null]);
10:52
<zcorpan>
since null isn't a Transferable and can't be converted to one
10:55
<Ms2ger>
What's the definition of postMessage?
10:59
<zcorpan>
void postMessage(any message, optional sequence<Transferable> transfer);
10:59
<zcorpan>
this is the worker postMessage
11:01
<Ms2ger>
Looks like it would throw, yes
11:03
<zcorpan>
hmm. cross-document messaging is a webapps spec, isn't it? i guess webapps needs to join the github thing if they want the tests i submitted there :-)
11:04
<zcorpan>
maybe i should send an email to public-webapps-testing about that
11:04
<Ms2ger>
We'll see what Art sorts out today
11:05
<zcorpan>
i guess the document structure in the testsuite is generated from the whatwg spec?
11:06
<zcorpan>
or maybe not
11:06
<Ms2ger>
darobin did it
11:07
<tobie>
zcorpan: if you want to use the tests I sent you, I'll need to run them through fb legal. Should be fast, but I still have to do it.
11:07
<Ms2ger>
Fun
11:07
<tobie>
Ms2ger: absolutely.
11:16
<darobin>
zcorpan: the structure is generated from the spec, err, I forgot which variant I used but the differences are small
11:17
<darobin>
as for webapps moving to github, everyone seems to be waiting on the approval from someone else
11:17
<darobin>
so what I would recommend is: JFDI
11:17
<darobin>
ask for forgiveness, not permission and all that
11:17
<darobin>
and even then, screw forgiveness
11:17
<darobin>
just do it
11:25
<Ms2ger>
darobin, Art said he'd try to sort it out today, so maybe we could wait for that :)
11:26
<darobin>
you're such a bureaucratic square Ms2ger
11:27
<Ms2ger>
If you wanted to jfdi, should've done that ages ago :)
11:28
<darobin>
Ms2ger: I'm not working on WebApps testing :p
11:29
<Ms2ger>
Why not? :)
11:31
<jgraham>
Because then plh will apply "jfdi" to his fingers with a hammer?
11:33
<darobin>
Ms2ger: because I have two or three other things to do...
12:12
<annevk>
Ms2ger: but progress events is a useless spec, might be worth adding :)
12:12
<Ms2ger>
True :)
12:12
<Ms2ger>
But he didn't ask that
12:14
<jgraham>
We could write the null spec with 0 conformance requirements so that we have one with complete test coverage
12:14
<jgraham>
Although by "write"
12:14
<jgraham>
I really mean "use HTML4"
12:16
<annevk>
brb, adding quotes to my desk
12:27
<annevk>
http://krijnhoetmer.nl/irc-logs/whatwg/20130315#l-281 :-(
12:27
<annevk>
Sounds like Microsoft's CSS tests
12:28
<annevk>
But then of course they do contribute, whereas others just don't care that much...
12:57
<tobie>
Ms2ger: meeting with Ecma next week. Seems like this is fixable.
12:57
<Ms2ger>
We'll see
12:58
<tobie>
oh come on. A bit of optimism wouldn't hurt. :)
12:59
<Ms2ger>
You must be new here :)
13:00
<gsnedders>
"this"?
13:00
gsnedders
wonders if tobie is trying to let others submit tests to test262
13:00
<Ms2ger>
Right
13:00
<tobie>
Ms2ger: you know I am. :)
13:00
<gsnedders>
I got the impression this had to go to the Ecma GA to get approved, and nobody had the drive to push it through there.
13:02
<jgraham>
gsnedders: You can tell we are talking javascript if you aren't sure what "this" referes to
13:02
<tobie>
jgraham: I was trying to find a joke to make around this but failed.
13:02
<Ms2ger>
Nicely done
13:03
<gsnedders>
But I know what this refers to! thisArg!
13:05
<zcorpan>
hmm. i'm trying to build cssom with the latest xref/ but get an error: ValueError: Expecting property name: line 26 column 1 (char 791)
13:05
<gsnedders>
How come WebKit/Gecko support blobs for WebWorkers?
13:06
<zcorpan>
File "/System/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/json/decoder.py", line 382, in raw_decode
13:06
<zcorpan>
obj, end = self.scan_once(s, idx)
13:06
<Ms2ger>
What's the deepest anolis frame?
13:06
<gsnedders>
I mean for the Worker constructor, here.
13:07
<zcorpan>
Ms2ger: the above is at the bottom of the traceback
13:07
<gsnedders>
zcorpan: But it's not a frame with Anolis in it.
13:08
<zcorpan>
ah. File "build/bdist.macosx-10.8-intel/egg/anolislib/processes/xspecxref.py", line 56, in buildReferences
13:08
<jgraham>
Is the latest xref/ actually valid json?
13:09
<Ms2ger>
Apparently not
13:09
<zcorpan>
what's broken?
13:09
<Ms2ger>
https://github.com/whatwg/xref/commit/dd82cfef17a9f9230fd487106a7475048f5ee79a
13:09
<Ms2ger>
You can't have a comma at the end of the line
13:10
<zcorpan>
oh, oops!
13:14
<jgraham>
Yeah, what a ridiculous notion
13:15
<zcorpan>
now make gives me `No such file or directory: 'data/xrefs/css/cssom.json'. Please create it first. ` and `[Errno 20] Not a directory: u'data/specs.json'`
13:15
<zcorpan>
though cssom.json exists there afaict :-/
13:17
<Ms2ger>
Oh, so
13:17
<Ms2ger>
Hm
13:18
<Ms2ger>
Maybe try --xref="../xref"?
13:20
<zcorpan>
Ms2ger: where?
13:21
<Ms2ger>
In the makefile
13:21
<Ms2ger>
Like https://github.com/whatwg/xhr/blob/master/Makefile
13:22
<zcorpan>
https://dvcs.w3.org/hg/csswg/file/654203ca5641/cssom/Makefile is the current makefile
13:24
Ms2ger
tries
13:25
<zcorpan>
i added --xref="data" (since that's where xref lives) but got the same result
13:34
Ms2ger
is confused
13:36
<zcorpan>
Ms2ger: about what?
13:53
<Ms2ger>
About the fact that anolis doesn't throw here
13:53
<Ms2ger>
But instead gives me an empty Overview.html
13:54
<Ms2ger>
The node stuff, probably
13:54
<Ms2ger>
Ah, here we are
13:56
<Ms2ger>
zcorpan, where do you get the webidl parser?
13:56
<zcorpan>
Ms2ger: https://dvcs.w3.org/hg/csswg/file/654203ca5641/cssom/README
13:59
<annevk>
Ms2ger: did you write that
13:59
Ms2ger
gives up
13:59
<annevk>
I mean zcorpan
13:59
<zcorpan>
annevk: no
13:59
<annevk>
Defining HTML inline with IDL seems like a terrible idea
13:59
<annevk>
That's what ReSpec does and every spec with ReSpec runs into problems because of it
14:00
<jgraham>
"defining html inline with idl"?
14:03
<zcorpan>
example idl file: https://dvcs.w3.org/hg/csswg/file/654203ca5641/cssom/idl/CSSCharsetRule.idl and the html source which references idls: https://dvcs.w3.org/hg/csswg/file/654203ca5641/cssom/cssom-source
14:03
jgraham
wonders why kris and plh aren't "ext" in Art's email
14:05
<annevk>
it's the sole thing I dislike about Art
14:05
<annevk>
his email client
14:05
<annevk>
so annoying, I think sometimes it has leaked into my contacts
14:06
<jgraham>
Well
14:06
<jgraham>
To be fair
14:06
<Ms2ger>
Hrm
14:07
<Ms2ger>
It claims to be thunderbird
14:07
<jgraham>
It's better than all the non-plain-text, gmail-toting, Googlers
14:07
<Ms2ger>
And I can confirm that it's not a feature of thunderbird
14:07
<jgraham>
I assumed that it was a feature of nokia
14:07
<jgraham>
That people without @nokia.com addresses are marked external
14:08
<jgraham>
But that theory seems to be wrong
14:08
<Ms2ger>
Also better than: howcome's indented >'s for quoting, Chris Lilley' 'INITIALS>'
14:09
<jgraham>
And <bryan>
14:10
<Ms2ger>
My response is in red
14:12
<tobie>
jgraham: :D
14:14
<annevk>
I'm glad I got you guys inspired, this is entertaining
14:20
<zcorpan>
just do what bjoern does and paper over people's email client's quirks
14:21
<Ms2ger>
I'd love to see his stuff implemented in my client
14:44
<zcorpan>
Ms2ger: jgraham: does https://github.com/w3c/web-platform-tests/pull/38 have a critic review? (how do i create one myself if there isn't?)
14:44
<zewt>
pretty much the most hopelessly broken thing in email today is people who do the "no quote markers at all" thing
14:45
<Ms2ger>
zcorpan, there is one
14:45
<zewt>
which is essentially "i don't care about how the internet does things and it doesn't bother me if i break the thread for everyone else"
14:45
<Ms2ger>
zcorpan, http://critic.hoppipolla.co.uk/r/8
14:46
<jgraham>
zcorpan: Did you add yourself as a reviewer on that repo?
14:46
<zcorpan>
no
14:47
<jgraham>
You should :)
14:47
<jgraham>
I will try to teach critic to comment on the pull request itself at some point
14:47
<zcorpan>
done. yeah that would be nice
14:50
<tobie>
yes. We'll be logging comments shortly, so having critic working properly there would be useful.
14:52
<jgraham>
Logging comments?
14:52
<tobie>
see Im hopeless after 3pm
14:52
<jgraham>
(I just meant critic would push a message saying "Critic review at {url}")
14:52
<tobie>
oh.
14:52
<tobie>
Well that's not very useful.
14:53
<jgraham>
(doing something more is probably possible but I'm not really sure hwo well the data models match)
14:53
<jgraham>
(e.g. as far as I can tell, github things of comments as things that apply to commits)
14:53
<tobie>
yeah, possibly.
14:53
<tobie>
some of them.
14:53
<jgraham>
(not to commits,pullrequest tuples)
14:54
<jgraham>
(also, less seriously, critic allows multiline comments and has two types of comment, whereas gh only has a single type of single line comment)
14:54
<jgraham>
*thinks
14:54
<tobie>
Don't make me have this conversation again.
14:55
<jgraham>
I'm not making you have any conversation
14:55
<jgraham>
I'm just pointing out that there are some difficulties
14:55
tobie
stays put.
14:56
<tobie>
:D
14:57
<jgraham>
http://developer.github.com/v3/pulls/comments/#create-a-comment does look a little more promising than I thought
15:00
<Ms2ger>
I complained on IRC once...
15:02
<Ms2ger>
Anyone who's on the webgl mailing list?
15:30
<zcorpan>
nice weekend guys
15:30
<annevk>
Hixie: http://ln.hixie.ch/?start=1086158925&order=-1&count=1 seems we are now doing those kind of OS-level APIs :-)
15:31
<annevk>
zcorpan: you too!
15:32
<zcorpan>
thanks :-)
15:59
<darobin>
heh
16:00
<darobin>
what actually happened: "If three of the most bitter rivals in the marketplace — all of whom have extensive experience in the subject in question — agree on something, then it's probably [...]"
16:00
<darobin>
A) that they don't want the competition
16:00
<darobin>
B) that they have their heads so far up their arses doing what they've done that they can't see alternatives
16:01
<darobin>
C) that they're on their way to lose their dominance or even disappear
16:01
<darobin>
D) all of the above
16:01
<SimonSapin>
darobin: Is that about mobile carriers in France?
16:02
<TabAtkins>
jgraham: The way to answer that question is. "Yes. Are you volunteering to write them? ^_^" (the smilie at the end is important)
16:06
<darobin>
SimonSapin: no, it's about the 2004 Hixie blog post that annevk pointed to
16:06
<darobin>
featuring Sun Microsystems, assuming you're old enough to remember who that is
16:07
<SimonSapin>
were they W3C members?
16:08
<annevk>
I've not had time to fix many bugs yet, but it's up at http://fetch.spec.whatwg.org/ now
16:10
<annevk>
I'm also thinking of putting basic CORS and redirect handling directly into basic fetch
16:11
<annevk>
Thoughts welcome
16:15
<darobin>
SimonSapin: oh yes
16:22
<SimonSapin>
darobin: I know about Sun, but it ceased to be before I knew about working groups
16:42
<darobin>
SimonSapin: they were very big in XML groups
16:43
<darobin>
and towards the end in anything related to mobile
17:12
<annevk>
darobin: depending on this bus I'm about to take, I might be a bit early or late
17:13
<darobin>
annevk: that's okay, we're heading out in a few minutes so we'll be there
17:14
<annevk>
darobin: are you close to Leicester Square now by any chance?
17:14
<darobin>
not really
17:14
<darobin>
I'm in Bethnal Green
17:15
<darobin>
I'll be walking to that pub
17:17
<annevk>
nice
17:17
<Ms2ger>
Enjoy
17:19
<smaug____>
scheib: ping
17:22
<darobin>
annevk: actually, I'm in the wrong TZ — I'll be walking there in about 30min
17:23
<annevk>
right
17:23
<annevk>
it'll take me about an hour anyway
17:29
<marcosc>
darobin: happy birthday, fella!
17:31
<darobin>
marcosc: thanks man!
17:36
<jgraham>
Did anne really just say "my arrival time depends on how long it takes me to get there"?
17:37
<jgraham>
TabAtkins: In this case it wasn't so mcuh "write them" as "rewrite them to not be part of the WebGL testsuite"
17:37
<jgraham>
But as it turns out zcorpan is awesome
17:38
<jgraham>
Next week: sky blue
17:57
<darobin>
"I don't always go to Bethnal Green, but when I do my arrival time depends on how long it takes me to get there" --annevk
19:04
<scheib>
smug____: at lunch. Email?
19:08
<Ms2ger>
smaug____, you mean
19:14
<smaug____>
scheib: ping me after lunch ?
19:15
<smaug____>
scheib: I'm wondering about the focus handling and pointer lock
19:15
<smaug____>
and spec is unclear about documents and windows
19:54
<scheib>
smaug____: Let's discuss pointer lock and focus handling.
19:55
<smaug____>
scheib: so, first, "Pointer lock must succeed only if the target is in the DOM tree."
19:55
<smaug____>
what if the target is in a data document
19:55
<smaug____>
say, document.implementation.createHTMLDocument().body.requestPointerLock()
19:55
<smaug____>
what should happen?
19:55
<scheib>
I'm unfamiliar why that wouldn't be considered as part of the DOM, apologies.
19:56
<Ms2ger>
It's part of the DOM
19:56
<Ms2ger>
But there's no browsing context
19:56
<smaug____>
yes, it is part of DOM
19:56
<smaug____>
right
19:57
<smaug____>
then, about focusing. What should happen to pointer lock when focus is moved to an iframe ?
19:59
<smaug____>
per spec pointer lock is cancelled in that case
19:59
<smaug____>
but I'm not sure whether implementations actually do that always
20:00
<scheib>
So, I'm not personally familiar with createHTMLDocument() yet, just started looking. It seems to create a new document for you, but is it rooted anywhere?
20:00
<smaug____>
"rooted"?
20:00
<smaug____>
it is just a document
20:00
<Ms2ger>
Not in a frame or somehting
20:01
<scheib>
Ok, so you can request a lock on a document, and it will be rejected from locking if it is not the document of an active tab, and the tab in the OS level active window.
20:01
<smaug____>
yeah, something like that. Need to just use spec terminology
20:01
<scheib>
BTW, if we can improve the wording of the specification I'm happy to do so, suggestions welcome.
20:01
<smaug____>
browsing context and such
20:01
<smaug____>
scheib: the focus handling is more interesting
20:02
<smaug____>
hmm, actually, per spec, focusing an iframe doesn't exit lock state, since the parent window is still focused... I think
20:02
<smaug____>
but that may mean that key events go to domain X yet mouse events to domain Y
20:02
<scheib>
Ok, on to that. If focus is in the main document, and not an iframe, I expect a request for pointer lock to fail if made from the iframe.
20:02
<Ms2ger>
"Every bug is two bugs: the bug in your code, and the test you didn't write"
20:03
<smaug____>
scheib: request pointer lock in top level document
20:03
<smaug____>
scheib: then focus iframe
20:03
<smaug____>
what should happen to pointer lock
20:03
<smaug____>
note, mouse isn't used to focus the iframe
20:03
<smaug____>
only .focus() or some such
20:04
<scheib>
Right. I don't see a reason it should exit (yet).
20:04
<smaug____>
well, it is really really odd if key events go to domain X yet all the mouse events to domain Y
20:04
<smaug____>
key events follow focus
20:04
<smaug____>
mouse events follow pointer lock
20:07
<scheib>
'odd' may be subjective here. If mouse input has been locked to a particular target I believe it is reasonable for it to stay locked to that target unless there is a reason to break it (user exit, app exit, user moves operating system focus away from the web page).
20:08
<scheib>
The ability to move the keyboard input focus is intentional, it should move, you should still be able to type into different input boxes, move around with TAB, etc.
20:08
<smaug____>
even cross domain?
20:08
<smaug____>
anyhow, the spec should clarify this
20:09
<scheib>
Crossing into an iframe, from a user point of view, is essentially the same as moving between regions of a web page. Unless we are protecting a user from a security vulnerability I don't know that they should be concerned about the domain, in the same way they aren't normally concerned when they move keyboard focus when outside of pointer lock.
20:10
<scheib>
Re clarifying the spec: certainly, we can do that, let's wrap up that we're in agreement as to what should happen.
20:10
<smaug____>
I think I can live with focus moving
20:10
<smaug____>
at least until a security bug is found
20:10
<smaug____>
:)
20:12
<scheib>
Ok. BTW, I verify that webkit allows the transition in current implementation via http://scheib.github.com/HTMLMisc/PointerLockAndFullscreen.html
20:13
<smaug____>
based on code inspection Gecko allows focus changes in fullscreen mode, but I was thinking to change it
20:13
<smaug____>
but maybe I don't do that yet
20:13
<smaug____>
I'm adding the non-fullscreen mode
20:20
<scheib>
In clarifying the spec for focus, how does this addition in section 7. Requirements sound to you: "The Pointer Lock API must exit the pointer lock state if the user agent, window, or tab loses focus." by adding, " Moving focus between elements in a tab, including into other browser contexts inside iframes, does not exit pointer lock."
20:21
<smaug____>
scheib: sounds ok
20:21
<smaug____>
thanks
20:36
<smaug____>
scheib: if pointer lock and focus are first in iframe, and then focus is moved to parent, what should happen?
20:36
<smaug____>
iframe isn't focused anymore, so I guess pointer lock is cancelled
20:46
<scheib>
smaug____: Same arguments as above, I don't see a reason to exit pointer lock.
20:47
<scheib>
E.g. use case, a 3D model editor that is a composition of 3D window views and a property editor on the right hand side. Pointer lock engaged inside one the of 3D views that are each an iframe, and then keyboard focus moved to a number edit box in the property editor in the parent document. Why break the lock?
20:53
<smaug____>
scheib: well, then the spec needs still some clarifications
20:54
<scheib>
" Moving focus between elements in a tab, including into other browser contexts across iframe boundaries, does not exit pointer lock." ?
20:55
<smaug____>
hmm, actually, need to drop iframe
20:55
<smaug____>
we have frames and what not
21:01
<scheib>
" Moving focus between elements in a tab, including into other browsing contexts, does not exit pointer lock." ?
21:07
<smaug____>
ok
21:07
<scheib>
smaug____: Any suggestions on how to clarify in specification terms the relationship between a document and a tab with active window manager focus?
21:17
<scheib>
smaug____: mmm, I'm thinking along the lines of "active document of the browsing context currently focused by the OS/window manger, e.g. in the active tab of a browser." http://www.whatwg.org/specs/web-apps/current-work/multipage/browsers.html#active-document
21:20
<smaug____>
yeah
21:20
<smaug____>
something like that
21:36
<Hixie>
zewt: <a download> exists now
21:37
<zewt>
that doesn't let me set a filename without also forcing a download (like content-disposition: inline; filename=foo)
21:37
<Hixie>
ah, right
21:38
<zewt>
don't think i ever really understood why we didn't do <a download filename=foo>, keeping them orthogonal (like they are in C-D)
21:39
<Hixie>
more complicated
21:41
<zewt>
keeping unrelated things separate seems simpler to me, at least
21:47
<zewt>
uh gross, firefox is recommending putting BOMs on files (in the console)
22:44
<smaug____>
Web Audio API spec is so sad