01:22
<Domenic>
tobie: what's the status on getting https://github.com/tobie/specref/pull/250 re-landed? What broke in prod, and can I help fix it?
11:34
<nox>
TabAtkins: Is element.matches("> *") supposed to be valid?
11:37
<nox>
TabAtkins: Seems like that method didn't get updated in DOM to use "Parse A Relative Selector" instead of "Parse a Selector".
13:48
<tobie>
Domenic: Some aliasing is broken somewhere. I'd need to think it through again and make sure I'm around to check the logs when we push it.
13:59
<annevk>
philipj: around? Any more thoughts on https://github.com/whatwg/html/pull/695?
14:39
<Domenic>
tobie: hmm I guess I was hoping I could stand up a server locally and debug failures
14:47
<tobie>
Domenic: you could of course totally do that. Don't know what I was thinking.
14:47
<tobie>
Domenic: I obv have a bit much on my plate atm.
14:47
gsnedders
buys tobie a bigger plate
14:49
<tobie>
gsnedders: ty
14:49
<gsnedders>
tobie: I'm always here looking for useful suggestions
14:50
<tobie>
Domenic: I can't remember the precise issues I was having, and the logger I use doesn't keep more than 24hrs of logs in its free version
14:50
<tobie>
Domenic: it had to do with some alises being broken
14:51
<Domenic>
tobie: so e.g. browser around the site and watch the logs for errors with the word "alias" in them>
14:51
<Domenic>
?
14:51
<tobie>
Domenic: possibly because of calling the search API used by specref.org
14:51
<Domenic>
hmmm
14:52
<tobie>
well, just run node index.js with whatever old version of node this thing requires
14:52
<tobie>
then just curl the different APIs for stuff related to the changes
14:52
<tobie>
Domenic: iirc it failed spectacularly and pretty quickly too
14:53
<Domenic>
philipj: how can you not think the warning helps for sync XHR given this trendline? https://www.chromestatus.com/metrics/feature/timeline/popularity/465
14:54
<Domenic>
philipj: at this rate it should be at 0 in 1.5 more years, literally!
14:55
<gsnedders>
Domenic: I can't imagine it dying till Fetch because practically usable
14:55
<annevk>
that alone shows how much the warning is worth
14:56
<annevk>
so much less synchronous IO
14:57
<tobie>
when was the warning introduced?
15:02
<annevk>
Quite a while ago, and every now and then someone gets upset with it and tells me how wrong I am and that for their particular niche synchronous IO is justified
15:03
<annevk>
Heck, even Brendan wants me to remove the warning
15:04
<gsnedders>
imo the big advantage of sync IO is the reduction of boilerplate with XHR
15:05
<tobie>
so why does this graphic prove that the warning has had an impact?
15:05
<annevk>
tobie: https://lists.w3.org/Archives/Public/public-webapps/2014JanMar/thread.html#msg232
15:06
<tobie>
All I'm seeing is a initial spike which is probably when the fist measures where taken, followed by a regular gradual decline.
15:06
<tobie>
To prove the warning works, you'd need a measurable acceleration in decline from the point at which it was introduced, no?
15:07
<annevk>
tobie: why?
15:07
<gsnedders>
there's been plenty of other evangelisation about the badness of sync IO
15:08
<gsnedders>
hence I'm with tobie here, the causation hasn't been shown
15:08
<tobie>
gsnedders: precisely.
15:09
<annevk>
I guess that's fair
15:10
<annevk>
I've seen plenty of anecdotal evidence that this warning helps folks decide, but I guess you can't really tell from that graph per se
15:19
<Domenic>
It certainly isn't hurting
15:26
<jyasskin>
mounir: Have you figured out which version of Tidy you're running yet?
15:26
<jyasskin>
It's frustrating to have https://github.com/w3c/permissions/blob/gh-pages/contributing.md tell me to use it, and then have patches rejected because I did.
15:35
<annevk>
Domenic: I care quite strongly about moving towards "then"
15:35
<annevk>
Domenic: 😃
15:35
<annevk>
Domenic: was there anything else that I missed due to it being hidden or can I land?
15:35
<Domenic>
annevk: I'd like to do a second pass later today, sorry
15:36
<annevk>
Domenic: sure, no rush
15:46
<annevk>
We're at a low point for outstanding PRs btw, been a while
16:02
<Domenic>
yeah it warms my heart
16:59
<annevk>
"195 bugs found." \o/
16:59
<annevk>
Of course, we now have well over 200 issues...
17:12
<philipj>
Domenic: I don't doubt that it has *some* effect, but there are non-deprecated things that are also slowly decreasing and I've never been able to see a clear connection between deprecation date and change in usage for anything else, so I just don't know what explains the decreasing usage
17:13
<philipj>
Whatever is driving the decrease, I'd expect it to show some kind of half-life decay and not linear decrease, so getting down to 0.01% or could take a very long time.
17:13
<philipj>
So, I think something more aggressive will be needed to get anywhere with this in a reasonable time frame
17:14
<philipj>
Maybe setting a timeout that's continuously decreased so that only 0.01% of page loads have a synx XHR timeout because of it, so that it's less and less reliable
17:15
<Domenic>
philipj: you don't buy my trendline argument that we only need 1.5 more years? :)
17:15
<philipj>
Domenic: oh, didn't realize you were bein tongue in cheek :)
17:18
<philipj>
one could also imagine an intervention where certain user interactions cause a timeout, but that might not be fun to debug for developers
17:22
<gsnedders>
philipj: you're still at Opera, right?
17:24
<annevk>
Is SharedWorker still not supported outside Window? https://www.w3.org/Bugs/Public/show_bug.cgi?id=28504
17:25
<annevk>
TabAtkins: where is https://www.w3.org/Bugs/Public/show_bug.cgi?id=28080 tracked on the CSS side?
17:27
<TabAtkins>
nox: I'd think .matches() is *supposed* to use an absolute selector. Where's the context it would be applying a relative selector to? (It's not the element; that's checked against the resulting subjects of the selector.)
17:28
<gsnedders>
philipj: nvm, realised what I need probably wasn't on t anyway
17:29
<TabAtkins>
annevk: Hm, I dropped the ball on that, sorry. I'll bring it back up.
17:42
<philipj>
gnarf: yep!
17:43
<philipj>
I mean gsnedders ^
17:43
<philipj>
gsnedders: I assume you know about presto-testo though?
17:45
<annevk>
Domenic: does your blocked element thing supersede https://www.w3.org/Bugs/Public/show_bug.cgi?id=23960?
17:46
<Domenic>
annevk: not really. Although the delegatesFocus stuff for shadow DOM might help a lot.
17:48
<annevk>
slightlyoff: do you know if anyone from Google is working on an undomanager? https://www.w3.org/Bugs/Public/show_bug.cgi?id=14337#c14
17:48
<gsnedders>
philipj: yeah
17:49
<gsnedders>
philipj: tl;dr: was thinking I wanted stuff from t/resources, actually need stuff from spartan-server
17:54
<philipj>
gsnedders: I'm not sure that's around any longer
17:57
<gsnedders>
philipj: oh, it will be
18:01
<annevk>
Domenic: also see https://www.w3.org/Bugs/Public/show_bug.cgi?id=24718 and https://www.w3.org/Bugs/Public/show_bug.cgi?id=24720
18:12
<Domenic>
annevk: what replaced "in a composed document"?
18:13
<annevk>
Domenic: in a shadow-including document
18:13
<annevk>
Domenic: which in turn is defined in terms of shadow-including root
18:13
<annevk>
Domenic: both already in DOM
18:14
<Domenic>
\o/
18:19
<nox>
TabAtkins: There are varions tests with such selectors un WPT, hence the question.
18:19
<TabAtkins>
That sounds like a bug in the tests!
18:31
<Domenic>
annevk: it would be minorly helpful if you could define "connected"
18:32
<annevk>
Domenic: you mean the IDL attribute?
18:32
<Domenic>
annevk: nah just the concept. But presumably you'd do both at the same time.
18:32
<annevk>
Domenic: isn't the concept just "in a shadow-including document"?
18:33
<Domenic>
annevk: hmm I guess so. It would be nicer if the author-facing API matched the concepts.
18:35
<annevk>
Domenic: I can still rename that easily
18:39
<gsnedders>
TabAtkins: wait you have the power to necro stuff? shit.
18:39
<TabAtkins>
I am a great and powerful sorceror.
18:54
<nox>
TabAtkins: Ok.
20:45
<smaug____>
annevk: Domenic: has it been written down somewhere that boolean attributes should be avoided in APIs unless it is absolutely clear what true/false mean in that context?
20:45
smaug____
is looking at directory-upload spec, and proposes using dictionary for the recursive param, not a boolean
20:46
<Domenic>
smaug____: it's in Web IDL
20:47
<Domenic>
smaug____: oh, no, I am misremembering. Web IDL just warns against booleans defaulting to true.
20:47
<Domenic>
smaug____: http://ariya.ofilabs.com/2011/08/hall-of-api-shame-boolean-trap.html is my favorite reference, but maybe we should add it to https://w3ctag.github.io/design-principles/
20:47
<TabAtkins>
yes plz
20:49
<smaug____>
Domenic: thanks, http://ariya.ofilabs.com/2011/08/hall-of-api-shame-boolean-trap.html is a good one
21:45
<tobie>
+1
21:45
<tobie>
TabAtkins: just curious if you use boilerplate beyond what bikeshed offers for your CSS specs.
21:45
<TabAtkins>
In what way?
21:45
<tobie>
TabAtkins: I'm not starting to work on spec'ing the concrete sensors (Gyroscope, AmbientLight, etc.)
21:46
<tobie>
TabAtkins: and there's a lot of stuff that are going to be common to each, even though I've stuck as much of it as I could in the generic sensor spec.
21:48
<tobie>
TabAtkins: for example, they'll all need to define the following: https://w3c.github.io/sensors/#definition-reqs
21:48
<TabAtkins>
Okay, gotcha.
21:48
<TabAtkins>
No, I don't do anything else. Dont' need to - the existing CSSWG boilerplate is plenty.
21:48
<TabAtkins>
But if you need such, there are a few ways to do it.
21:49
<tobie>
TabAtkins: well, not only do I think it's worthwhile for me, it might also be something nice to offer other editor who want to expose sensors
21:49
<TabAtkins>
Most obvious is to set up some new boilerplate for yourself, derived from the dap boilerplate. Adjust whatever's necessary.
21:51
<tobie>
TabAtkins: sure, but that's lightly different, though. I feel like I need scaffolding more than boilerplate
21:51
<tobie>
*slightly
21:51
<TabAtkins>
Another way, probably better for your needs, is to build an inclusion file, and just include that into your specs via a <pre class=include> (which I realized a few days ago wasn't documented).
21:51
<TabAtkins>
Hmm, scaffolding is *way* more difficult.
21:51
<TabAtkins>
Well, actually, maybe not. Depends on how complex you mean.
21:51
<tobie>
I'm not sure, yet, actually. :-/
21:51
<TabAtkins>
So, again this isn't documented yet because I'm a bad person, so bear with me, when including a file you can set up some custom macros that will be available inside the included file.
21:51
<TabAtkins>
So if it's it's just a matter of "here's a template, fill in the holes", then that's totally possible.
21:52
<tobie>
Maybe that's what I'm looking for.
21:52
<TabAtkins>
If it's any more diverse, tho, then there's really nothing that can be done mechanically, and the right thing is just to document what each specific sensor is expected to provide.
21:53
<tobie>
Sure. But it would be pretty cool if I could write the specifics of each as YAML or JSON or whatnot and have a whole bunch of things generated.
21:54
<TabAtkins>
Like I said, if it's just a matter of a template with holes that will be filled with text, I can do that. (If the holes are *large* it's not *convenient* currently, but can be done.)
21:54
<TabAtkins>
Any more complex and I'd have to invent something new, and I'm wary to try and do that without more examples to inform the design.
21:55
<tobie>
oh, I wasn't asking for work on your side
21:56
<tobie>
Just curious as to what can be done with the existing tool and in which direction I should poke
21:56
<tobie>
I'll start off mostly manual with the first spec
21:58
<tobie>
then try to extract commonalities from it.
21:58
<tobie>
Thanks for the pointers
21:59
<TabAtkins>
np
22:47
<slightlyoff>
annevk: I don't. I think that stopped when rniwa went to apple