06:39
<annevk>
Eric Portis (he/him): it currently does, in https://github.com/annevk/orb/issues/30 I mention it as a possible fast path, but to match its processing model we'd have to track it across redirects
06:41
<annevk>
Eric Portis (he/him): I think if we do it properly (i.e., account for redirects) it's probably reasonable to safelist such responses
10:33
<Ondřej Žára>
Looks like the "Libera" IRC is not the proper whatwg chat place, right?
10:34
<Ondřej Žára>
That might explain noone responding to my messages ;-)
10:42
<annevk>
Ondřej Žára: I don't think we ever advertised anything but Freenode and this channel, and now it's just this channel
10:43
<Ondřej Žára>
Ondřej Žára: I don't think we ever advertised anything but Freenode and this channel, and now it's just this channel
Yeah, I have no memories about how I got from Freenode to Libera. Whatever.
10:43
<Ondřej Žára>
Let me repaste my original Libera question then
10:43
<Ondřej Žára>
are there any general guidelines regarding "js property <-> HTML attribute" reflection for custom elements?
10:43
<Ondřej Žára>
for instance those interactive components whose properties change quickly (<my-map lat=... lon=...>)
10:43
<Ondřej Žára>
I recall people suggesting to use attributes as the source of truth
10:43
<Ondřej Žára>
But that might not hold any more?
10:52
<annevk>
Might be a good question for #web-components:matrix.org though it's not very active. There's definitely no guidelines in the standard. I'd recommend looking at https://html.spec.whatwg.org/#reflect and seeing how it might be applicable for a custom element. At least if you want them to behave like built-in elements. Often that does use the attribute as a source of truth, but you could optimize around that a fair bit.
10:53
<Ondřej Žára>
Thanks for the pointers. Will read the spec part a bit, then probably re-ask in that channel.
13:58
<Eric Portis (he/him)>
annevk: Ok, thanks! This would make the story for third party media hosts that want to upgrade resources without requiring every document that embeds from them to also upgrade, well, livable, I think.
14:49
<annevk>
Eric Portis (he/him): I mean in principle ORB shouldn't really impact anyone, though I suppose there might be some perf implications if you don't have your MIME types in order
14:50
<annevk>
Eric Portis (he/him): it might be a reasonable way out however for new image formats and over the long term require something like that for all cross-origin fetches
16:03
<smaug>
Domenic: will be there in the meeting real soon
17:59
<sefeng>
Noam Rosenthal: Just want to make sure I understand the rationale behind the LargestContentfulPaint spec. So if the image is expanded, we don't report the actual painted size, but rather we report the intrinsic size of the image
17:59
<sefeng>
is it because we don't want web devs to game the metric?
18:06
<Eric Portis (he/him)>

annevk: “in principle ORB shouldn't really impact anyone”

Half of all <img>s load cross-origin; I’d suspect CSS background-image to have a similar breakdown. My mental model of ORB previously was: in order to adopt new image formats, lots of HTML was going to have to change, and there just wouldn’t be a way to use new image formats across origins in CSS. Big impact! Mitigated by CORP.

18:08
<annevk>
Eric Portis (he/him): I see, new media formats definitely have that wrinkle atm
18:08
<annevk>
Eric Portis (he/him): I'd be okay with this compromise
18:08
<annevk>
Eric Portis (he/him): do you want to put it forward in https://github.com/annevk/orb/issues/3?
18:09
<Eric Portis (he/him)>
annevk: will do!
18:15
<Noam Rosenthal>
Noam Rosenthal: Just want to make sure I understand the rationale behind the LargestContentfulPaint spec. So if the image is expanded, we don't report the actual painted size, but rather we report the intrinsic size of the image
That’s the basic idea. We don’t want an enormously scaled 1x1 image to be considered “largest”
18:18
<sefeng>
Noam Rosenthal: I see, I start to understand the penalty factor thing..but if a scaled 1x1 image is the largest, what's wrong of reporting it..?
18:30
<Noam Rosenthal>
It’s ok but you don’t want to report it as 1000x1000
18:43
<sefeng>
Noam Rosenthal: why don't we if that's what gets painted?