00:00
<Hixie>
typically they have them already, since their objects will typically be in arrays they're walking over
00:00
<TabAtkins>
Yeah.
00:00
<Hixie>
but in any case generating IDs is trivial
00:00
<TabAtkins>
Note that there is a use-case for wanting to get the set of all objects with a given point inside them.
00:00
<Hixie>
yeah that's a harder problem
00:01
<TabAtkins>
That one is generally going to be slow unless you have custom solutions.
00:01
<TabAtkins>
Like doing searches on a k-tree or something.
00:01
<Hixie>
the a11y use case and the normal click hit testing use case don't need it, so i'm not too worried about it
00:01
<TabAtkins>
Yeah, can be added later.
00:01
<TabAtkins>
Without impacting the top-object use-case.
00:05
<Hixie>
if we add a way to addHitRegion() a path giving that region a label to read out when an AT brings the AT cursor to that part of the canvas
00:05
<Hixie>
should we just make fillText() and strokeText() always imply such a call?
00:06
<TabAtkins>
With a path equal to the bounding box of the text or something?
00:06
<Hixie>
yeah
00:06
<Hixie>
or something
00:06
<TabAtkins>
Off the top of my head, I don't see why not. Seems nice.
00:06
<Hixie>
(not bounding box given fillText() along a path)
00:06
<TabAtkins>
Yeah, sure.
00:06
<TabAtkins>
Suitably mutated was-a-rectangle-before-following-the-path.
00:06
<Hixie>
it wouldn't survive a drawImage() from one canvas to another, probably, which would be unfortunate
00:07
<TabAtkins>
*none* of the hit-testing would.
00:07
<Hixie>
right
00:07
<TabAtkins>
You need retained-mode if you want that.
00:07
<Hixie>
also would probably not survive drawing text onto a path then drawing the path
00:07
<TabAtkins>
Hm. That's not necessarily ture.
00:07
<TabAtkins>
UAs could retain the information alongside the path.
00:08
<Hixie>
and would be kinda annoying if you were to draw many layers of one piece of text, e.g. doing a motion blur effect
00:08
<TabAtkins>
True.
00:08
<Hixie>
actually that one is probably the most serious
00:09
<Hixie>
also it would make it annoying to use fillText() to draw pictograms and things from custom fonts
00:09
<Hixie>
hmm
00:09
<Hixie>
maybe best to leave it to explicit calls
00:09
<TabAtkins>
Hm. Yeah, might want something explicit.
00:09
<TabAtkins>
Though extending *Text to make it easier might be good too.
00:09
<TabAtkins>
Just not by default.
00:10
<Hixie>
add an optional ariaRole argument, maybe
00:10
<Hixie>
(what's the appropriate aria role for a graph label, anyone know?)
00:10
<TabAtkins>
Is there a good role for just 'text' or 'label' or something?
00:10
<TabAtkins>
Heh, lemme check.
00:11
<Hixie>
looks like there's neither
00:11
<Hixie>
<label> defaults to No Role
00:11
<Hixie>
that's unfortunate
00:12
<TabAtkins>
Yes, reading the list of them in /TR/wai-aria/roles, I see nothing appropriate.
00:13
<Hixie>
ok. i'm gonna go to tgif, and will pick up this canvas stuff again next week.
00:14
<TabAtkins>
Lemme know if anything cool happens - we're doing board games today.
08:28
<Onderhond>
Quick question: is it just me or was there some kind of proposition for a <template> element in html5?
09:31
<annevk>
Onderhond: there is
09:35
<annevk>
sigh Microsoft does not want to remove constants from Indexed DB?
09:35
<annevk>
consider it for v2?
09:35
<annevk>
euh
09:42
<annevk>
oh lol
09:42
<annevk>
drm, no drm, it's like https versus http
09:43
<annevk>
public-html turning into a comedy list does seem positive
09:44
<Ms2ger>
Better than a tragedy?
09:45
<annevk>
well, public-html is not good tragedy, so yes
10:16
<Onderhond>
annevk: where can I read up on the <template> element? Google isn't very helpful.
10:17
<Ms2ger>
XBL2 or Web Components.
10:17
<Ms2ger>
?
10:25
<hsivonen>
I wonder why Gmail thinks emails from Mark Watson (@netflix.com) would be spam if I didn't have a filter telling Gmail never to mark them as spam
11:00
<Dashiva>
hsivonen: I think you can hover over some part of it to see why it was marked
11:00
<Dashiva>
Or maybe that's priority inbox
11:39
<annevk>
fyi https://plus.google.com/107429617152575897589/posts/NZBJe6Jjt1f
12:07
<annevk>
there's a <!-- XXX WTF? --> in DOM4
12:07
<annevk>
about the traverse siblings algorithm it seems
12:09
<Velmont>
annevk: Bah. They did not :S -- Well, they can actually change the numeric constants and leave the constants as-is. I guess their developers don't rely on 1 and 2, but rather IDBTransaction.READ_WRITE etc.
12:09
<Ms2ger>
Hah
12:09
<Ms2ger>
I would not count on such a reasonable thing
12:10
<Velmont>
Ms2ger: Well, I don't think Microsoft would admit that they have such sloppy developers.
12:10
<annevk>
Velmont: but you know... you rewrote their tests
12:11
<Velmont>
annevk: Heh, -- they didn't use the numbers there, they used the constant names.
12:11
<annevk>
ah, at least they got something right :)
12:12
<Ms2ger>
Have you looked at the WebStorage tests?
12:13
<annevk>
"Interestingly enough, the IETF follows the snapshot model, deals with hundreds of highly critical standards and it works absolutely well."
12:13
<Velmont>
Not really, took a quick peek once.
12:13
<annevk>
hmm
12:13
<annevk>
HTTP - latest stable version, out of date, a mess
12:13
<annevk>
data URLs - a mess
12:13
<annevk>
URI/IRI - cannot be implemented as is
12:14
<annevk>
TLS - not updated with respect to what implementations actually need to do
12:14
<annevk>
HTTP authentication -- hahaha
12:14
<hsivonen>
I wonder how chuck@jumis manages to confuse server-side login with DRM and Trusted Computing for DRM and Trusted Computing against janitor attacks
12:15
<hsivonen>
did I get trolled?
12:15
<annevk>
MIME - maybe this one is too easy
12:15
<Ms2ger>
Man, I wish WebGL used testharness.js
12:16
<annevk>
hsivonen: I find his emails usually confusing
12:42
<gsnedders>
Ms2ger: Likewise. WebKit's test harness really doesn't work if you want actual regression tracking…
12:43
<Ms2ger>
Oh, is this all WebKit's fault?
12:43
<gsnedders>
Well, at least it looks exactly like WebKit's JS test framework, so I was assuming it was.
12:46
<gsnedders>
Also: I don't get how people don't bang heads against people claiming random crash bugs are security issues.
12:49
<gsnedders>
And, ah, how I've missed dealing with GWT output.
12:55
<Ms2ger>
So, what's the current thinking about http://dev.w3.org/2009/dap/file-system/file-writer.html ?
12:59
<asmodai>
hsivonen: Happen to know if FF fires events for the discarding of large images on a tab switch?
13:02
<Ms2ger>
To the page? No
13:02
<asmodai>
is there any other way to detect it?
13:02
<Ms2ger>
Shouldn't be
13:02
<Ms2ger>
You could try a timing attack
13:09
<hsivonen>
asmodai: no idea
13:10
<hsivonen>
annevk: regarding indexed db, clearly, there's no v1, either, as evidenced by whar Isreal said
13:12
<annevk>
well yeah, there's no v
13:12
<hsivonen>
asmodai: what's the use case?
13:13
<asmodai>
hsivonen: Was asking based on a tweet of a buddy of mine, I think he's working on his framework debugger again
13:13
<asmodai>
hsivonen: Trying to get his ass in here so he can just ask himself XD
13:14
<annevk>
maybe it's time for a new "red pill" post
13:14
<annevk>
on standards
13:15
<hsivonen>
annevk: go for it
13:20
<hsivonen>
XML ER is not on xml-dev yet
13:52
<pflock>
Does anyone know why the lines here all have the same color though I gave them different strokeStyle's? http://jsfiddle.net/u4Gpf/
13:54
<Philip`>
pflock: You need to call context.beginPath() before each new path
13:54
<pflock>
ah thanks ;)
13:54
<Philip`>
Otherwise you're drawing a path with 1 line, then extending the path to 2 lines and drawing all that again, etc
13:55
<pflock>
yes I noticed that he always colored them all in the last color I set, so that one overwrites the other colors, but I didn't know why
13:55
<pflock>
I thought context.stroke() already 'ends' the path
14:04
<annevk>
ok so far I've got
14:04
<annevk>
* living
14:05
<annevk>
* public domain
14:05
<annevk>
maybe developed in public is newish enough too
14:06
<Ms2ger>
* no telcons
14:06
<annevk>
ah yeah, asynchronous public decision making
14:06
<annevk>
nice one
14:08
<mhausenblas>
maybe add implementation-driven?
14:09
<Ms2ger>
Or in close cooperation with implementations, at least
14:10
<mhausenblas>
in this respect I'm rather radical, Ms2ger
14:10
<Ms2ger>
I'm not :)
14:10
<mhausenblas>
what's the point if it isn't implemented?
14:10
<mhausenblas>
why?
14:10
<Ms2ger>
Well, sure, it needs to be implemented
14:10
<mhausenblas>
implemented and implementable
14:11
<mhausenblas>
right
14:11
<mhausenblas>
I remember in horror the work in MPEG-7
14:11
<Ms2ger>
But it sounded like you were saying it had to be implemented before starting on a spec
14:11
<mhausenblas>
I was young and I needed the job
14:11
<mhausenblas>
mhm, sorta
14:11
<mhausenblas>
like impl - spec -impl - spec
14:11
<mhausenblas>
...
14:12
<mhausenblas>
of course you don't know the entire space so it must be iterative
14:12
<mhausenblas>
dunno who said it but something along the line of 'any complex system evolved from a very simple on'
14:13
<Ms2ger>
Mm
14:13
<mhausenblas>
so: KISS(at least) in the beginning and iterate
14:13
<mhausenblas>
s/simple on/simple one
14:13
<Ms2ger>
Agreed
14:20
<annevk>
I don't think implementor involvement is important to mention
14:20
<annevk>
if they're not involved
14:20
<annevk>
nothing happens
14:22
<annevk>
can anyone do a quick review? http://annevankesteren.nl/2012/02/standards-red-pill
14:23
<mhausenblas>
annevk I don't understand your logic
14:24
<mhausenblas>
in what you say it looks implementer involvment is an essential precondition, no?
14:24
<annevk>
what part?
14:24
<annevk>
yeah
14:24
<mhausenblas>
so how can it be *not* important to mention
14:24
<mhausenblas>
people often make the mistake to forget to externalise the most basic assumption
14:25
<annevk>
mkay
14:25
<mhausenblas>
because they seem to obvious
14:25
<mhausenblas>
right
14:25
<Philip`>
annevk: s/proof/prove/
14:25
<mhausenblas>
annevk, lemme know when you're done - happy to (re)tweet/G+ it
14:26
<annevk>
thanks to Philip` and your addition I guess it's done now
14:26
<mhausenblas>
cool
15:24
<kennyluck>
Language — US English is the standard
17:51
<jonnybarnes>
hey, has anyone got a good refrence list of XML references?
17:51
<StoneCypher>
google does
17:52
<jonnybarnes>
Looking at an RSS feed I've made with PHP and seeing a lot of &#13; entities and trying to find out what character thats meant to be
17:52
<Ms2ger>
CR
17:52
<StoneCypher>
so use an entity decoder
17:52
<jonnybarnes>
Cheers Ms2ger
17:52
<StoneCypher>
or give a man a fish
17:53
<jonnybarnes>
StoneCypher: any recomendations for entoty decoders?
17:53
Ms2ger
slaps StoneCypher around a bit with a fish
17:53
<jonnybarnes>
*entity
17:53
<StoneCypher>
jonnybarnes: google does
18:08
<Philip`>
jonnybarnes: http://www.fileformat.info/info/unicode/char/search.htm?q=%26%2313%3B&preview=entity should do it for numeric ones
18:08
<Philip`>
(&#13; means decimal 13, which is hex 000D, so it's Unicode character U+000D)
18:11
<jonnybarnes>
cheers Philip`
20:03
<hsivonen>
cute how DRM proponents don't like my use of the word adversary even though I used it because the proposal used that word
20:19
<StoneCypher>
they're adversaries of adversary?
20:42
smaug____
is glad he isn't following public-html
20:42
<smaug____>
Commenting "Encrypted Media proposal" would make me too angry
20:42
<Ms2ger>
I'm always glad I'm not following public-html
21:35
<Hixie>
wtf
21:35
<Hixie>
i have a sqlite table
21:36
<Hixie>
it has a column whose values are all TRUE
21:36
<Hixie>
i do a select on that table with a GROUP BY clause, and select the aforementioned column
21:36
<Hixie>
the result set has that column with the value NULL
21:39
<Velmont>
grrrrreat! Sure they're all TRUE?
21:39
<Hixie>
yes
21:39
<Hixie>
i've never set them, and i only just added the column with default true
21:40
<Velmont>
A+ stuff. I'd guess either a very strange behaviour or a good old bug. -- From what I've heard about SQLite it might actually be the first one...
21:44
<Hixie>
hm, i wonder what it's value actually is
21:44
<Hixie>
it returns TRUE
21:44
<Hixie>
but sqlite doesn't have a TRUE value...
21:44
<Hixie>
ok i replaced whatever values it had with 1 and now it works
21:44
Hixie
decides to move on in blissful ignorance
23:50
<gsnedders>
Object.defineProperty(String.prototype, "foo", {get: function(){"use strict"; print(typeof this);}}); "foo".foo; --- why is that object, and not string? 10.4.3 passes thisArg along directly, after 8.7.1 has invoked [[Call]] with the primitive value…
23:50
<gsnedders>
V8 claims string, SM/Carakan claim object.
23:50
<gsnedders>
I'm claiming SM/Carakan bug…