07:36
<MikeSmith>
botie, inform annevk https://lost-contact.mit.edu/afs/cern.ch/w3.org/www/Consortium/Agreement/Full.html "Ownership of Copyrights and Patents" & "Licenses"
07:36
<botie>
will do
09:50
<darobin>
I thought I had seen a page on the wiki listing the conventions for ID names (e.g. for DOM stuff) but I can't seem to find it
09:50
<darobin>
did I dream it or just failing at search?
09:51
<darobin>
there's an incoming patch for much better WebIDL support in ReSpec and I'd like to use the opportunity to completely align with the convention; having a page to point the guy to would help
09:52
<darobin>
ah, found it
09:55
<Ms2ger>
Good
11:01
<zcorpan>
is it defined when :active starts and stops applying relative to events like mousedown, mouseup, click ?
11:03
<zcorpan>
http://software.hixie.ch/utilities/js/live-dom-viewer/saved/3341 :-/
11:13
<Ms2ger>
zcorpan, that sounds like a job for... D3E
11:13
Ms2ger
cackles
11:33
<zcorpan>
http://software.hixie.ch/utilities/js/live-dom-viewer/saved/3342 does what i expect in firefox but in my other test case it says 'false' for both events. can't figure out what the difference is
11:44
<botie>
annevk, at 2014-12-18 07:36 UTC, MikeSmith said: https://lost-contact.mit.edu/afs/cern.ch/w3.org/www/Consortium/Agreement/Full.html "Ownership of Copyrights and Patents" & "Licenses"
12:37
<zcorpan>
i kind of got stuck here. https://critic.hoppipolla.co.uk/r/3537
12:37
<zcorpan>
possibly i've made a stupid mistake and i don't see it
13:13
<zcorpan>
oh. i think i found it. e.preventDefault cancels :active from applying!
16:45
<gsnedders>
so jgraham, are we still trying to maintain a linear history in html5lib-python? because up till now we basically have.
16:46
<gsnedders>
even though it doesn't integrate well with github's PRs
16:46
<jgraham>
gsnedders: I think we should give up on that :(
16:47
<jgraham>
There are so many PRs that just need to be merged
16:47
<gsnedders>
jgraham: my problem is I don't like merging stuff till we have tests "passing" again
16:48
<gsnedders>
because it just becomes way more time consuming checking for regressions
16:49
<jgraham>
gsnedders: So last I saw the PR for adding an ignore list had "don't merge this" on it
16:49
<gsnedders>
jgraham: because of the questions in the thread
16:52
<gsnedders>
jgraham: i.e., how do we handle failing tests involving lone surrogates
16:52
<gsnedders>
jgraham: the other option is to just fix the bug and worry about this another time :P
16:55
<jgraham>
gsnedders: I am happy with that option
16:56
<gsnedders>
I'll /try/ and get something done this evening. This may well not happen.
19:28
<Hixie>
annevk: you around? what's the story on intents these days?
20:15
<annevk>
Hixie: no real story
20:16
<annevk>
Hixie: we have some ideas, but limited bandwidth to experiment
20:16
<annevk>
Hixie: it seems like service worker to service worker communication will come first, perhaps that'll provide some way to let sites figure out client-side cooperation
20:17
<annevk>
Hixie: although that doesn't give you the decentralized stuff of course
20:17
<annevk>
ato: that's not just the case for Error's prototype
20:18
<annevk>
ato: all prototypes are global-bound
20:18
<annevk>
ato: (re tweet)
20:19
<ato>
annevk: RIght, I was trying to figure out why err instanceof Error was acting weird across chrome- and content contexts in XPCOM.
20:24
<annevk>
ato: you should also know that Gecko has a non-standard implementation of instanceof for "DOM" objects
20:24
<annevk>
ato: (unless that was put into IDL)
20:24
<annevk>
ato: Gecko's instanceof for "DOM" objects doesn't care about the window the object belongs to
20:27
<ato>
Interesting.
20:31
<roc>
annevk: I believe that's in the spec
20:32
<roc>
heycam definitely told me it's in the spec :-)
20:32
<annevk>
http://heycam.github.io/webidl/#es-interface-hasinstance
20:32
<annevk>
I'm not a fan, we already moved stuff from "DOM" to ES, this would break whenever we do that
20:33
<annevk>
E.g. this will break for promises in Gecko
20:35
<annevk>
https://twitter.com/arslancb/status/545674463530844161 foolip?
20:36
<roc>
is it really the case that, e.g., there is one Array prototype per window so that [] created in one window is not an instanceof Array in another window?
20:37
<annevk>
roc: afaik, yes
20:37
<roc>
if so, that's broken and needs to be fixed in ES
20:38
<annevk>
roc: Allen is not too interested :-(
20:38
<roc>
sucks to be him
20:38
<annevk>
heh
20:39
<annevk>
I doubt we could actually make them the same at this point
20:39
<roc>
why?
20:41
<Domenic>
wow, no, that is a super-importnat property of ES
20:41
<Domenic>
programmers depend on it a lot
20:42
<Domenic>
it's extremely useful to be able to mutate the prototype in one window and have it not affect another
20:42
<Domenic>
i was just doing that this morning in fact
20:43
<TabAtkins>
Yup, I've done it before too.
20:43
<TabAtkins>
Lets you lift the magic attached to some objects, like arrays, without polluting the "normal" array.
20:43
<TabAtkins>
Because subclassing doesn't work.
20:47
<roc>
well
20:47
<roc>
the behavior of instanceof is the important thing here
20:48
<roc>
we don't actually need the prototype objects to be the same
20:54
<Domenic>
... that is literally the definition of instanceof
20:54
<annevk>
Domenic: it has a hook now so objects can define instanceof to behave however they wish
20:54
<Domenic>
sure. let's say: that is literally the definition of instanceof that JS developers expect.
20:55
<roc>
unfortunately developers also expect an <img> to be an instanceof HTMLImageElement
20:55
<Domenic>
nah, not if it comes from another frame
20:56
<Domenic>
then it should be instanceof otherFrame.contentWindow.HTMLImageElement
20:56
<roc>
no-one expects that
20:56
<roc>
except a few standards gurus
20:56
<Domenic>
instanceof is not a nominal type test
20:56
<Domenic>
if you want the name just use tagName
20:56
<Domenic>
or .constructor.name or similar
20:58
<roc>
http://stackoverflow.com/questions/1760492/how-can-i-tell-if-a-javascript-object-is-an-image-or-a-canvas
21:14
<Hixie>
annevk: if i have someone who wants to contribute, where should i point them to? (for intents)
21:39
<annevk>
Hixie: I guess the WHATWG list and that thread on sharing is all there is thus far
21:39
<annevk>
Hixie: and the WHATWG wiki page on sharing
21:39
<Hixie>
k
23:06
<jgraham>
Domenic: [citation needed] on "that's what js developers expect". testharness.js has at least two places which say "we want to use instanceof here but can't because we don't know which window the object is from"
23:06
<TabAtkins>
Read "what js developers expect" as "what [sufficiently enough that we can't break backwards compat] js developers expect"..
23:06
<jgraham>
I believe that at least one of those was previously a bug
23:08
<jgraham>
TabAtkins: "this is an unfortunate legacy behaviour we can't change" is totally different from "that's what developers expect".
23:08
<TabAtkins>
Well, I've come to expect it, at least.
23:09
<jgraham>
Sure but you have also come to expect document.all. It doesn't mean it's a great ideea
23:21
<Hixie>
http://w3c.github.io/web-animations/ has some messed up MathML stuff
23:21
<Hixie>
search for <{
23:22
<birtles>
Hixie: I'll look into it
23:22
<birtles>
... but I blame bikeshed already :)
23:25
<birtles>
TabAtkins: was there a recent change to bikeshed's handling of mathml?
23:25
<TabAtkins>
Shouldn't have been.
23:25
<birtles>
my local copy of bikeshed works fine but tip seems to be broken
23:25
<TabAtkins>
BUT WHO KNOWS
23:25
<TabAtkins>
(I didn't do anything on purpose.)
23:26
<TabAtkins>
Ahahaha, I know what happened.
23:26
<TabAtkins>
I wrote a custom serializer, and lxml's handling of namespaces is hilariously bad. I'll file an issue and deal with this asap.
23:26
<birtles>
TabAtkins: thanks!
23:27
<Hixie>
oh man, lxml
23:27
<Hixie>
lxml is one of the reasons i dumped anolis :-)
23:29
<MikeSmith>
in the python world I'm not sure what practical alternative there is to lxml
23:30
<TabAtkins>
There really isn't, particularly since I use CSSSelect. :(
23:31
<MikeSmith>
yeah whatever other python things there might be, I think they have worse problems
23:31
<Hixie>
yeah, i dumped python for that same reason
23:31
<MikeSmith>
baby meet bathwater
23:32
<MikeSmith>
python has the advantage of being something a lot of other people have some familiarity with already
23:33
<MikeSmith>
and anyway in my experience at least lxml just works pretty well overall. It's got some quirks but again in my experience at least it's not a giant hardship to work around them
23:34
<MikeSmith>
I say all this as somebody who's neither really fond of lxml nor python myself
23:35
MikeSmith
goes back to fixing problems with a broken python script for web-platform-tests...
23:44
<MikeSmith>
does chown -r not follow symlinks?
23:45
<Hixie>
probably depends on -h ?
23:45
<Hixie>
dunno
23:47
<MikeSmith>
k
23:48
<MikeSmith>
hmm yeah it seems for sure it doesn't
23:48
<MikeSmith>
looking the man page and the docs for the -h and -H args
23:49
<MikeSmith>
even with -H, "If the -R option is specified, symbolic links on the command line are followed. (Symbolic links encountered in the tree traversal are not followed.)
23:49
<Domenic>
any ideas how to do *:not(.example *) in a browser that doesn't support complex selectors inside not?
23:49
<Hixie>
MikeSmith: ah well there you go
23:50
<Hixie>
Domenic: you want to match all elements that aren't a child of .example?
23:50
<Domenic>
Hixie: yep
23:50
<Hixie>
Domenic: :not(.example) > *, :not(.example) > * *
23:50
<Hixie>
hm, though that won't work
23:51
<Hixie>
for descendants
23:51
<TabAtkins>
No, can't do it without more constrained tree.
23:51
<TabAtkins>
(So that you can explicitly knock everything else out.)
23:51
<TabAtkins>
And even then it's terrible.
23:51
<Hixie>
Domenic: wait, i don't understand what you want here.
23:51
<Hixie>
Domenic: if you just want "not child of .example", then ":not(.example) > *"
23:51
<TabAtkins>
All element that aren't inside of a .example.
23:52
<Hixie>
is that what :not(.example *) does?
23:52
<Hixie>
so it's like the equivalent of my :not(:matches(.example #)) ?
23:52
<Domenic>
I think :not(.example) * might work
23:52
<Domenic>
So basically I have this CSS reset `* { box-sizing: border-box; margin: 0; padding: 0; }`
23:53
<Domenic>
but I noticed it was also affecting my examples
23:53
<Domenic>
which is bad
23:53
<TabAtkins>
Domenic: Nope. That selector <.example><foo><div>
23:53
<Hixie>
":not(.example) *" will match... what TabAtkins said
23:53
<TabAtkins>
Rather, that selects the <div> there.
23:53
<Domenic>
ah
23:53
<TabAtkins>
Hell, it'll match <.example><div> as long as <.example> has an ancestor.
23:53
<Hixie>
the old fashioned way to do this is:
23:53
<Hixie>
* { ... }
23:53
<Hixie>
.example * { ... }
23:53
<Domenic>
also its specificity is way higher than * so that's causing my own problems
23:53
<Domenic>
yeah
23:53
<TabAtkins>
It'll only rule out elements which have .example as *all* their ancestors.
23:54
<Domenic>
Hixie: yeah I tried that but I can't get the margin/padding to go back to their "defaults"
23:54
<Hixie>
well you can hard-code the defaults
23:54
<Hixie>
40px for blockquote margin-left, etc
23:55
<Domenic>
it may be time to just do a targeted margin: 0/padding: 0 on all the things I am using
23:55
<Domenic>
Wish Chrome supported complex :not's
23:57
<Domenic>
While I'm hear does anyone know what's the deal with `-webkit-margin-before` and friends?
23:57
<TabAtkins>
what about them?
23:58
<Domenic>
what do they even do, why do they need nonstandard properties
23:58
<TabAtkins>
They're like margin-left, but logical directions rather than physical.,
23:58
<TabAtkins>
And they're non-standard because we haven't defined logical margins yet.
23:58
<Domenic>
logical = related to right-to-left/left-to-right?
23:58
<heycam>
TabAtkins, if you could do that soon it would be good :-)
23:58
<TabAtkins>
Yes, direction and writing-mode.
23:58
<heycam>
I'm fixing up our logical properties right now
23:58
<TabAtkins>
heycam: Yeah, fantasai and I intend to. :/
23:59
<Domenic>
cool, thanks
23:59
<heycam>
TabAtkins, one thing that's not clear from the various mailing list postings is whether logical properties are exposed on a computed style object