06:51
<annevk>
FYI: I plan on landing the CORB PR at https://github.com/whatwg/fetch/pull/686 later today or tomorrow
11:24
<annevk>
mkwst: you around?
11:24
<annevk>
mkwst: what's the status on https://www.w3.org/TR/clear-site-data/?
11:25
<annevk>
mkwst: we're interested in implementing, but I recall Chrome had some perf issue and I wonder if that's reflected in the draft yet
11:25
<mkwst>
I am not around. Will be out of meetings around 17:00.
11:26
<mkwst>
We had to turn off `cache` for some period of time because there was some framework that added `clear-site-data: cache` to every response, which crashed Chrome.
11:26
<annevk>
ta
11:26
<mkwst>
AFAIK, it's working again after we did some work to make sure that deletions were not parallelizable.
11:27
<mkwst>
But I'd have to check with my colleague who did that work.
11:27
<annevk>
Okay, so just need to be careful to avoid DOS
11:28
<mkwst>
Right. I think we got into trouble because we were double-, triple-, and quadruple-deleting cache entries, which our implementation turned out not to like.
11:28
<mkwst>
So don't do that.
11:29
<annevk>
baku will try 😃
13:26
<annevk>
Domenic: so sed is reasonably fast to process source, but it's super awkward to deal with the newlines of the large <details> block
13:26
<annevk>
Domenic: it ends up looking rather ugly
13:27
<annevk>
Domenic: and I don't know enough bash to make it look better
13:27
<wanderview>
annevk: perl!
13:27
<Domenic>
Heh. Yeah we could do some research, or switch to Python or Node
13:28
<annevk>
wanderview: I'd like HTML build scripts to no longer rely on Perl (or Pascal)
13:29
<wanderview>
annevk: ok... i was just saying that with the right flags perl is pretty close to a sed drop-in and it supports newlines much better
13:29
<wanderview>
but... whatevs
13:32
<Domenic>
Oh, I thought it was a joke... Person 1: "my code is too ugly"; person 2: "you should use perl!"
13:35
<annevk>
heh
13:38
<gsnedders>
Domenic: hah, Perl is good for some things! :)
13:52
<edent>
I'd like to make a PR to the HTML spec. Is there a guide for how to set up BikeShed (?) so I can see if my changes have worked?
13:55
<gsnedders>
edent: see https://github.com/whatwg/html-build
13:57
<annevk>
edent: and https://github.com/whatwg/html/blob/master/CONTRIBUTING.md
14:22
<edent>
Perfect, thanks!
16:48
<TabAtkins>
devsnek: As annevk said, yes, but currently it's just a flag calling out to implementors and users that it *should* have that behavior; it doesn't actually *do* anything on its own, so the spec author still has to manually write the caching behavior. (Thus annevk's mention of [Cached], which will actually have some semantics.)
16:49
<devsnek>
good to know, thanks
18:44
<smaug____>
has there been any discussion to be able to load custom elements definitions lazily?
18:44
<smaug____>
Basically a notification when a new CE-like element (without definition for it) is created
18:45
<smaug____>
then one could fetch the definition, and upgrade
18:46
<smaug____>
(FWIW, this is something that Firefox UI code seems to need while replacing XBL with CE, but FF UI code can do script loading synchronously in these cases, I think)
18:48
<Domenic>
We briefly discussed that at the last F2F and said people should use MutationObservers, I thought?
18:48
<smaug____>
oh, indeed we did discuss :)
18:50
<smaug____>
but, wasn't there some requirement for localName filtering
18:50
<smaug____>
https://github.com/whatwg/dom/issues/398
20:16
<nox>
annevk: Argh.
20:17
<nox>
annevk: Just realised that `for (let foo in …)` loops in Servo with the old SM are completely broken, and closures in the loop end up sharing the same foo value. :(
21:50
<Domenic>
Servo is a legacy browser, news at 11. (I kid because I love.)
22:11
<gsnedders>
nox: have you considered upgrading SM? ;P