00:02
<jamesr__>
chromium is big. web platform is very broad
00:02
<caitp->
yes, and yet i manage to get 7 minute clobber builds of m-c on my laptop
00:03
<caitp->
oh the things we miss
00:08
<jamesr__>
what kinda laptop?
00:08
<jamesr__>
i can clobber build chromium in about 3 minutes, but it takes a lot more power than a single laptop
00:10
<jgraham>
I think it's pretty community spirited of Google to make building Chromium have such a high barrier to entry
00:11
<jgraham>
It makes Gecko look appealing in comparison
00:17
<jamesr__>
i'm pretty sure nobody chooses to make the codebase harder to work with. in fact, quite a few people do the opposite. it's hard
00:18
<jgraham>
I imagine it's less pressing when you can build in 3 minutes due to a huge internal build cluster
00:26
<jsbell>
Ms2ger: https://critic.hoppipolla.co.uk/r/3713
00:26
<jsbell>
dang, not online... but cc'd, good 'nuff
06:07
<hemanth>
meow
06:07
hemanth
paws at catip, annevk
06:59
<Domenic>
enough roleplaying in whatwg, I think...
07:09
<hemanth>
Ooops, Sorry Domenic. (It's become a (bad) habit)
08:06
<annevk>
morning
08:09
<terinjokes>
annevk: you're supposed to roleplay
08:11
<annevk>
heh
08:15
<hemanth>
^_^
10:05
<annevk>
this may interest some here: https://annevankesteren.nl/2015/01/javascript-web-platform
10:29
<hemanth>
annevk: internal states, tell me about that! ;)
10:33
<annevk>
hemanth: hah, still unclear?
10:33
<annevk>
hemanth: in the JavaScript specification it is fairly explicit, internal slots are the [[thingies]]
10:34
<annevk>
hemanth: in web platform land you have to read a bit, e.g. https://xhr.spec.whatwg.org/#formdata has an [[entries]] slot and an internal "create an entry" method
10:35
<hemanth>
annevk: not that i'm not clear, but finding those [[ ]] in MediaElement ain't easy (for me)
10:36
<Ms2ger>
I wonder if it'd be clearer if rather than prose "represents" or "has associated" requirements, we'd have a table with all the slots and maybe their type
10:36
<hemanth>
agree, need to "read a bit" :)
10:36
<annevk>
Ms2ger: that's my internal slots IDL proposal
10:36
<annevk>
hemanth: pretty much anything with an attribute getter requires an internal slot
10:37
<Ms2ger>
You mean, do it in the IDL block?
10:37
<annevk>
Ms2ger: yeah, the idea is that basically everything that's an attribute automatically creates an internal slot
10:37
<Ms2ger>
Eh, not sure I like that
10:37
<annevk>
Ms2ger: unless you say it doesn't need one (e.g. innerHTML doesn't need one, we'd generate the string on the fly)
10:37
<hemanth>
annevk: okies...
10:38
<Ms2ger>
We'd have to opt out all the reflecting attributes too
10:38
<annevk>
yeah, not everyone does
10:38
<annevk>
Ms2ger: I think for reflecting attributes we want special IDL
10:38
<annevk>
Ms2ger: [Reflect=URL] or some such
10:38
<Ms2ger>
And with my implementor hat on, I'm not sure I like having it intertwined with the interface definition
10:38
<Ms2ger>
I don't know
10:38
<hemanth>
annevk: https://twitter.com/GNUmanth/status/535356806856331264 is not possible due to internal slots, right?
10:38
<annevk>
Ms2ger: Chrome has that, looks pretty neat
10:39
<annevk>
hemanth: I'm not sure what that is saying
10:40
<hemanth>
well annevk gnumanth in me ;) So...we can't just do an Object.observe(input, function(changes){})
10:41
<hemanth>
/in/is
10:41
<Ms2ger>
That's what you have with accessors
10:42
<annevk>
hemanth: ah yes, that doesn't work automatically because of getters rather than actual properties
10:42
<hemanth>
#sad
10:43
<hemanth>
anyway MutationObserver can come to rescue, right?
10:43
<annevk>
hemanth: a thought I had was that if we formalize internal slots through IDL we could also formalize Object.observe as acting on those formalized slots
10:43
<hemanth>
that would be great! annevk
10:43
<annevk>
hemanth: mutation observers are only about tree state, not about object state
10:43
<hemanth>
oh :(
10:44
<hemanth>
annevk: you got a chance to read http://h3manth.com/new/blog/2015/es6-reflect-api/ ;) ?
10:45
<annevk>
hemanth: do now, though I read about Reflect before, I'm on es-discuss :-)
10:45
<hemanth>
:)
10:49
hemanth
runs, got late for the cab...
11:14
<annevk>
JakeA: I for one, cannot wait for the internal build of the pre-alpha
11:16
<JakeA>
annevk: haha, I'm soooo fucking angry
11:18
<annevk>
JakeA: shows that post-religion people are still gullible
11:55
<annevk>
For some reason www.w3.org no longer loads in my Firefox Nightly
11:56
<annevk>
Is Firefox Nightly trying to protect me?
12:18
<JakeA>
annevk: yesterday, when I was talking about vague error messages, I was referring to steps 6.2 and 7 https://w3c.github.io/push-api/#widl-PushManager-subscribe-Promise-PushSubscription
12:19
<JakeA>
They're trying to decide between "be vague, and allow push services the freedom to throw the best error" and "update the spec for new errors created by different kinds of push services"
12:20
<JakeA>
I prefer the former
12:20
<JakeA>
EME does the latter
12:21
<JakeA>
https://w3c.github.io/encrypted-media/#h3_methods-1 1.2.3
12:22
<annevk>
JakeA: you mean EME does the former and you prefer the latter?
12:23
<annevk>
JakeA: what EME does is a bug that they should fix
12:23
<annevk>
JakeA: can't really hand-wave the protocol like that :-(
12:23
<JakeA>
annevk: hah, yes that's what I mean
12:34
hemanth
likes jsbin and 6to5 integration :)
13:19
<iandevlin>
I feel dumb when I read anything in this channel :-o
13:20
<JakeA>
fwiw so do I
13:20
<JakeA>
I suppose it's better than thinking you know it all, right?
13:20
<iandevlin>
True
13:20
<iandevlin>
But I appear to know nothing :D
13:21
<iandevlin>
That said, I often struggle to open food packaging so hey.
14:14
<annevk>
iandevlin: I think that's how everyone in this channel started out
14:15
<annevk>
iandevlin: hard to say whether we "improved" over time by learning though :-P
14:16
<iandevlin>
annevk: not sure if that's encouraging or not!
14:18
<annevk>
iandevlin: one day I hope social sciences starts studying mailing lists and IRC channels
14:19
<iandevlin>
annevk: haha :-)
14:39
<MikeSmith>
https://www.drupal.org/node/2405903
14:39
<MikeSmith>
does Drupal actually emit <meta name="charset" charset="utf-8" />
14:40
<Ms2ger>
Lolwat
15:02
<MikeSmith>
hsivonen: https://html5.validator.nu/ and https://validator.nu/ returning 502
15:20
<zcorpan>
MikeSmith: the validator messages would be clearer if it reordered the attributes so that "charset" is first before checking against the RelaxNG
15:20
<zcorpan>
MikeSmith: i think that is similarly done for <input> to put type first
15:23
<MikeSmith>
zcorpan: the validator's not doing any reordering of attributes anywhere
15:24
<MikeSmith>
as far as I know
15:24
<zcorpan>
MikeSmith: i recall something about <input type> at least to make error messages less confusing
15:24
<zcorpan>
hsivonen: ^
15:26
<zcorpan>
MikeSmith: try this <input alt type=file> <input type=file alt>
15:26
MikeSmith
tries
15:26
<zcorpan>
and compare with <meta name=charset charset=utf-8> vs <meta charset=utf-8 name=charset>
15:28
<zcorpan>
without the reordering it would say that type=file is invalid for the first case
15:32
<MikeSmith>
OK, I see what you're saying but I don't think that's due any to the code doing any special reordering of attributes in that case -- instead it's just because of the way the RelaxNG schema defines the constraints for <input>, which is different from how it defines the constraints for <meta>
15:33
<MikeSmith>
because dealing with <input> is a massive PITA due to the fact it's not really one element but actually 17 or whatever different elements masquerading as one element
15:33
<annevk>
MikeSmith: at what point are you going to cut out the middle man?
15:36
<MikeSmith>
annevk: some day when I either suddenly get much smarter or when I suddenly have a bunch more time free to work on solving interesting problems
15:37
<Ms2ger>
zcorpan, fwiw, bz believes http://w3c-test.org/quirks-mode/table-cell-width-calculation.html flakiness in Gecko is because it assumes imgs load sync
15:38
<zcorpan>
Ms2ger: yes, that matches my analysis
15:38
<zcorpan>
Ms2ger: question is if we want <img src=data:...> to load sync or async
15:39
<Ms2ger>
I don't want anything :)
15:39
<annevk>
zcorpan: I want async...
15:40
<zcorpan>
MikeSmith: just need to double your intake of tea and then you can handle it yourself
15:40
<zcorpan>
annevk: why?
15:40
<MikeSmith>
heh
15:41
<MikeSmith>
zcorpan: it's imaginable that we could change the meta-checking user-visible behavior to be more like the input-checking user-visible behavior. But another thing is that the meta-checking behavior is itself also already relatively baroque, but just in a different way. Good times all around
15:41
<annevk>
zcorpan: consistency
15:41
<annevk>
zcorpan: though I guess <img src=known-uri> is also sync...
15:41
<annevk>
bah
15:41
<zcorpan>
MikeSmith: i guess <meta> invalidness might be rare and not worth optimizing for
15:42
<MikeSmith>
yeah true probably so
15:42
<zcorpan>
annevk: it is, yes
15:42
<zcorpan>
annevk: it loads sync in blink but not in current blink/webkit
15:43
<Ms2ger>
I think you blinked a blink there
15:44
<MikeSmith>
zcorpan: I think in general we could move more stuff out of the RelaxNG schema and into the Java code. Maybe <meta> is a specific case where that could happen, I dunno. But in general we get better behavior for end users the more we move stuff out that way when we can
15:44
<zcorpan>
gecko/webkit
15:45
<zcorpan>
MikeSmith: is there any data about which messages or elements or attributes are most common?
15:46
<MikeSmith>
nah we would need to add some stuff to the stats-collecting code to get that
15:46
<MikeSmith>
Hixie also talked to me about that recently
15:46
<MikeSmith>
so I should do it
15:47
<MikeSmith>
one nice thing about that is , it doesn't require me to be so smart
15:47
<zcorpan>
i think hsivonen did something like that in the early days
15:47
<MikeSmith>
yeah?
15:47
<MikeSmith>
if so I don't think he kept the code for collecting it in there
15:48
<MikeSmith>
anyway, it would be very useful info to have
16:04
<zcorpan>
MikeSmith: https://lists.w3.org/Archives/Public/public-html/2008Jan/0305.html
16:09
<zcorpan>
heh, "Bad value “X-UA-Compatible”" "Hmm."
16:10
<zcorpan>
does someone have IE? i can't get modern.ie to give me a screenshot for http://software.hixie.ch/utilities/js/live-dom-viewer/saved/3066
16:19
<iandevlin>
zcorpan: I have IE, what do you want?
16:20
<zcorpan>
iandevlin: can you tell me what the log says for the above link?
16:21
<iandevlin>
zcorpan: In IE11, the log says: log: 0
16:21
<iandevlin>
rendering mode: CSS1Compat
16:21
<iandevlin>
document has no title
16:21
<iandevlin>
oops
16:21
<zcorpan>
thanks iandevlin
16:21
<zcorpan>
means loading is async
16:24
<TabAtkins>
What does e10s expand to?
16:25
<jgraham>
eeeeeeeeees
16:25
<jgraham>
Or electrolysis
16:25
<jgraham>
Your choice
16:26
<iandevlin>
See? I don't even understand those logs that I pasted :-/
16:27
<Domenic>
https://lists.w3.org/Archives/Public/public-html/2008Jan/0305.html seems to basically just be saying "if people make an error often enough we should make it conforming" :-/
16:29
<Domenic>
Oh I see that's from 2008
16:30
<jgraham>
Domenic: It's more like "if people do X and we can't think of a reason that X is actively harmful (in some way that doesn't require everyone else to not do X too), then it should be conforming"
16:30
<Domenic>
Yeah, that doesn't sound like a conformance checker/validator too much to me. More like an interoperability checker.
16:30
<zcorpan>
iandevlin: annevk: Ms2ger: https://github.com/ResponsiveImagesCG/picture-element/issues/223#issuecomment-69772765
16:31
<annevk>
zcorpan: \o/
16:31
<annevk>
zcorpan: ah yes, allowing asynchronous decoding is a good one
16:31
<jgraham>
Domenic: Souns like one to me, just where the definition of "conformance" is based on data rather than someone's idea of aesthetics
16:31
<iandevlin>
zcorpan: aha thanks
16:31
<annevk>
zcorpan: anything sync is PITA, see about:blank
16:32
<Ms2ger>
Domenic, problem is that validators can't whine *too* much, otherwise people ignore them
16:32
<Domenic>
Ms2ger: yeah, I'm sympathetic to that. A lot of those seemed pretty bad though. e.g. border="0" instead of CSS.
16:33
<Ms2ger>
Domenic, *shrug*
16:33
<zcorpan>
Domenic: why is it better to use style="border:0" for something like a badge that you want other people to paste into their sites?
16:33
<Ms2ger>
Domenic, I mean, CSS is better and all, and I'd never use border="" today, but is it really harmful?
16:33
<jgraham>
Domenic: Right that makes sense if you can articulate a particular reason it's bad
16:34
<jgraham>
(some things that "everybody knows" are terrrible turn out to be not-so-terrible-after-all under this test)
16:34
<jgraham>
(e.g. <b>)
16:34
<zcorpan>
(border=0 might not be necessary at all today though since browsers dropped the default border on image links)
16:34
<Domenic>
zcorpan: knowing and propagating knowledge of the general-purpose mechanism is better than knowing that img's have a special-purpose attribute that will tweak the border.
16:36
<zcorpan>
Domenic: so presentational stuff is ok if it's general-purpose?
16:38
<Domenic>
not what i said
16:40
<annevk>
So nobody any idea why Firefox would refuse to load www.w3.org links?
16:41
<Ms2ger>
Doesn't for me
16:41
<Domenic>
is it on the HSTS preload list? There's some part of the w3 site that sets the HSTS header, but the rest of the site redirects https to http
16:47
<annevk>
Domenic: that sounds like it could be it
16:47
<zcorpan>
Domenic: yeah, sorry
16:48
<annevk>
Domenic: seems I cannot inspect items in the list in Firefox, I can either enable or disable it...
16:48
<annevk>
Domenic: but I do end up with infinite redirects...
16:48
annevk
didn't realize the W3C would try to set that header before they made things work
16:50
<annevk>
"Unfortunately these threads usually devolve into an endless thread of discussion and debate that become next to impossible to actually read through later without several hours of spare time and a bottle of whiskey..." hehehe
17:32
<MikeSmith>
annevk: is your Firefox refusing to load www.w3.org links right now?
17:33
<annevk>
MikeSmith: yes
17:33
<annevk>
"Firefox has detected that the server is redirecting the request for this address in a way that will never complete."
17:33
<MikeSmith>
ok yeah
17:33
<MikeSmith>
this happened to others
17:34
<MikeSmith>
including me, but in Chrome in my case
17:34
<annevk>
MikeSmith: the only way for w3.org to solve this is either no longer redirect https to http or set a HSTS header for everyone that removes the policy
17:34
<annevk>
MikeSmith: strict-transport-security: max-age=0 or some such
17:34
<MikeSmith>
did Tim or somebody send out public mail about this?
17:34
<annevk>
MikeSmith: I don't know
17:37
<annevk>
MikeSmith: yeah it's definitely HSTS
17:37
<MikeSmith>
yeah
17:52
<JakeA>
TabAtkins: Which spec defines a css property that gives an element an aspect ratio?
17:52
<JakeA>
(or am I making it up?)
17:52
<TabAtkins>
An old blog post of mine defined a bad version of it.
17:52
<TabAtkins>
But no spec does.
17:52
<JakeA>
No worries. Must have imagined it
20:05
<jsbell>
MikeSmith: can you verify that inexorabletash is a contributor to w3c/IndexedDB on github?
20:18
annevk
was reading that and thought, hey, I know inexorabletash, that's jsbell!
20:18
<annevk>
And then I read who wrote the line
20:18
<jsbell>
:)
20:30
<hsivonen>
MikeSmith, zcorpan: validator.nu is up again. thanks
21:03
<GPHemsley>
MikeSmith, jgraham: Yes, I was paraphrasing.
21:55
<Hixie>
so confused
21:55
<Hixie>
suppose i have an Iterator in JS
21:55
<Hixie>
do i use for (let a in g), for (let a of g), or for each (let a in g) ?
21:59
<jsbell>
Hixie: for (let a of g)
22:00
<Hixie>
ok, thakns
22:34
<TabAtkins>
Hixie: for(let a of g)
22:35
<TabAtkins>
"for each" was part of a very old proposal, only existed in Firefox if you invoked the right ritual version.
22:35
<TabAtkins>
Testing: ping?
22:39
<TabAtkins>
Ooof, I'm in netsplit right now. Awesome.
23:18
<MikeSmith>
jsbell: d'oh
23:18
<MikeSmith>
fixed now
23:20
<MikeSmith>
jsbell: I had created a github team odd the idxdb errors but has neglected to actually associate it with the idxdb repo
23:21
<MikeSmith>
*of the idxdb editors
23:25
<jsbell>
MikeSmith: np, thx
23:26
<jsbell>
MikeSmith: history pushed, you should be able to redirect dvcs.w3.org/hg/IndexedDB now
23:28
<MikeSmith>
jsbell: cool, will set it up in a couple hours
23:29
<MikeSmith>
(on from my phone now)