| 02:20 | <Kaiido> | How should issue sections be marked in bikeshed? There is one in Web IDL with just class="issue" that isn't styled appropriately and thus reads as a (weird) normative section. |
| 02:50 | <bakkot> | IIRC there's been some discussion in the past of having a method on AbortSignal that's like "register this handler if the signal is not already aborted, otherwise immediately call it", which is a pretty common use case (some discussion https://github.com/domfarolino/observable/issues/19#issuecomment-1698201158). anyone recall a discussion along those lines and where it could be found? |
| 02:50 | <bakkot> | I might just be thinking of throwIfAborted though. |
| 02:57 | <bakkot> | (note that this is basically how promise then works: "register this callback if the promise is not already settled, otherwise call it", modulo details about what "immediately" means) |
| 03:04 | <bakkot> | ah, node has util.aborted also, which is this but plus a weakref and in a promise form instead of registering a callback https://nodejs.org/api/util.html#utilabortedsignal-resource |
| 03:08 | <bakkot> | aha, which ultimately led me to https://github.com/whatwg/dom/issues/946 |
| 03:25 | <TabAtkins> | How should issue sections be marked in bikeshed? There is one in Web IDL with just |
| 03:32 | <TabAtkins> | Filed https://github.com/whatwg/meta/issues/289 |
| 06:52 | <annevk> | Kaiido: we use class=XXX |
| 08:11 | <annevk> | zcorpan: freddy: https://github.com/whatwg/html/issues/2249#issuecomment-1700567077 |
| 08:28 | <freddy> | I think we should do something once our URL parser is aligning on interop. What anne suggests sgtm.... zcorpan? |
| 08:29 | <zcorpan> | freddy: yeah I was just writing a comment |
| 08:37 | <freddy> | thanks! |
| 09:18 | <zcorpan> | annevk: thanks for the heads up! |
| 13:01 | <annevk> | zcorpan: it's an interesting question whether data: URLs count as cross-origin there; Fetch doesn't consider them as such, but HTML prolly does the quirks check and gets to decide and I doubt it special cases data: URLs |
| 13:02 | <annevk> | zcorpan: and it prolly shouldn't, so I guess I'd be inclined to treat it as a thing to fix in Chromium |
| 13:05 | <zcorpan> | annevk: "If the document has been set to quirks mode, has the same origin as the URL of the external resource" - the origin of the data: URL is an opaque origin, so is not same origin with the document |
| 13:07 | <zcorpan> | I can file a bug |
| 13:18 | <zcorpan> | Looks like Firefox also loads the stylesheet (but then doesn't hang because data: base urls are not allowed). So we need to fix this too |
| 13:27 | <annevk> | zcorpan: thanks, we should add this test too (perhaps the one that hangs Chromium with a -crash suffix?) |
| 13:29 | <zcorpan> | Isn't hang and crash different? |
| 13:29 | <annevk> | zcorpan: yeah, not sure how to test that in a useful way for Chromium then, maybe just have a short timeout or some such |
| 13:32 | <zcorpan> | annevk: I'd expect the result to be TIMEOUT, and the default timeout is probably short enough to not cause issues. It's slightly annoying when someones opens the test |
| 13:32 | <zcorpan> | annevk: we can also test the quirk origin check without hanging chromium, by checking computed style or so |
| 13:44 | <zcorpan> | https://bugs.chromium.org/p/chromium/issues/detail?id=1477672 |
| 13:44 | <annevk> | zcorpan: makes sense, since it just hangs the tab I'd create two tests, but it would be nice to cover that scenario so other browsers don't run into it accidentally |
| 13:57 | <zcorpan> | annevk: curious that firefox passes the test Parsing: <#x> against data:,. I thought data: wasn't allowed as base |
| 13:59 | <zcorpan> | is hash-only always allowed? |
| 14:02 | <zcorpan> | https://bugzilla.mozilla.org/show_bug.cgi?id=1850965 |
| 14:03 | <zcorpan> | fyi freddy |
| 14:03 | <annevk> | zcorpan: yeah |