| 02:10 | <bradleymeck> | I'm guessing there is no way to change the cache key for type=module's module map at this point, but I was wondering why it was chosen to be the request url. that maps to --perserve-symlinks approximately in node which was attempted to be the default and then backed out due to problems |
| 02:11 | <bradleymeck> | I've heard some claims of optimizations being doable on the request url based cache but haven't managed to dig up talks about them |
| 02:18 | <domfarolino> | Domenic: The Console wpt PRs very well may be on your post-vacation todo list but just pinging here in case they got buried; they might be some low-hanging fruit :) |
| 02:20 | <Domenic> | domfarolino: oh, nope, totally forgot, thanks for the ping! |
| 02:20 | <domfarolino> | π |
| 02:20 | wanderview | did not know you could have avatars in irc(cloud)... |
| 05:13 | <Domenic> | Sometimes I wonder if ekr ever expresses any positive thoughts about any changes to the web platform |
| 05:13 | <Domenic> | You'd think stricter content-type checking would be a slam dunk, but no... |
| 05:16 | <annevk> | Domenic: where is this? |
| 05:17 | <Domenic> | https://github.com/mozilla/standards-positions/issues/81#issuecomment-380328972 in particular, mozilla/standards-positions in general |
| 05:17 | <annevk> | Domenic: I think ekr would be pretty positive on TLS, but since he's very involved I'd expect him to be critical of that too π |
| 05:19 | <Domenic> | Also some hurtful words from Martin Thomson, but I guess that's just people being mean on the internet. |
| 05:19 | <Domenic> | s/mean/inconsiderate/ |
| 05:19 | <annevk> | Yeah, though I agree that it's hard to digest |
| 05:20 | <annevk> | And I actually had a wrong take on the whole thing initially so fortunately no PRs landed yet and JakeA chimed in |
| 05:28 | <annevk> | Concern from ekr seems valid though, esp for the sniffing bits |
| 05:28 | <annevk> | I agree that some more Content-Type enforcement might be good, but this also introduces new complexity (such as a new response type) to do it |
| 05:33 | <Domenic> | I thought they were changing that to just network error |
| 06:03 | <JakeA> | Domenic: but that would be observable via fetch(). I think it needs to return an empty response instead. |
| 06:05 | <JakeA> | Domenic: I'm proposing the same thing for a similar issue https://github.com/whatwg/fetch/pull/560/files#comment-180417935 |
| 06:07 | <Domenic> | Well that was the original proposal |
| 06:08 | <Domenic> | But to align with existing nosniff it could just be a network error if we think we could get away with it |
| 06:09 | <bradleymeck_> | After some time, and being unable to sleep due to it. I think I want to suggest changing import.meta.url to be on the request URL not the response URL |
| 06:09 | <bradleymeck_> | I guess making a new issue on just HTML would be the way to do that? |
| 06:23 | <annevk> | Domenic: unfortunately it does not quite align and the other bits would require a different factoring too |
| 06:23 | <annevk> | bradleymeck_: I guess, but does not seem acceptable |
| 06:23 | <Domenic> | bradleymeck_: you can open the issue, but we won't change. It makes it useless for its primary use case. |
| 06:24 | <bradleymeck_> | I claim that use case analysis is flawed after being unable to sleep thinking on it |
| 06:24 | <Domenic> | I claim knowing the module map key is not a use case we care for on the web |
| 06:25 | <bradleymeck_> | I don't think the use case is just around that |
| 06:25 | <bradleymeck_> | I think using the Response has mental model problems with `import()` using `import.meta.url` |
| 06:25 | <bradleymeck_> | mostly due to how the module map works |
| 06:25 | <annevk> | I recommend getting some sleep first |
| 06:26 | <Domenic> | import() also resolves relative to the response URL fwiw |
| 06:26 | <bradleymeck_> | well this is a can of worms if true |
| 06:26 | <bradleymeck_> | since all of this conflicts with node |
| 06:26 | <bradleymeck_> | and we can't reenable the --perserve-symlinks approach that this resembles due to resource usage issues |
| 06:26 | <Domenic> | Yeah not really sure what node is thinking. |
| 06:27 | <bradleymeck_> | well it tried this approach |
| 06:27 | <Domenic> | But they can make their own decisions and balance priorities |
| 06:27 | <bradleymeck_> | Domenic: it tried this approach and backed away after problems |
| 06:27 | <bradleymeck_> | it isn't like it just does what it wants without thinking |
| 06:28 | <annevk> | To be clear, this approach has worked fine for CSS et al |
| 06:28 | <Domenic> | Sure that's why I said make their own decisions and balance priorities not randomly pick something. |
| 06:28 | <annevk> | The web has tested this long before Node |
| 06:28 | <bradleymeck_> | in CSS I'm not sure how to compare the extra load operations that this incurs |
| 06:29 | <bradleymeck_> | or maybe I'm misreading something about how resolve works |
| 06:29 | <Domenic> | (in case you weren't aware static import *also* resolves relative to the response URL) |
| 06:29 | <bradleymeck_> | still bad situation in my mind |
| 06:30 | <annevk> | And with shared workers |
| 06:30 | <bradleymeck_> | unsure how having multiple module records for a response URL is a good thing |
| 06:30 | <annevk> | Shared workers even has the request map thingy |
| 06:30 | <Domenic> | Anyway, I'm heading to sleep. Hope Node can figure something out, but the web has a solution it's happy with. |
| 06:31 | <bradleymeck_> | I'm glad this talk has been as unwelcoming to my hopes as the last 12+months |
| 06:31 | <bradleymeck_> | anywho seems like no point in interacting |
| 06:33 | <annevk> | bradleymeck_: Iβd be interested in seeing how the mismatch might end up affecting developers, but the response URL being authoritative is a long established fact on the web |
| 06:34 | <bradleymeck_> | annevk: if the module map was response URL based I wouldn't think it a problem |
| 06:37 | <Domenic> | Oooh before I fell asleep I had a hunch and it was ima |
| 06:38 | <Domenic> | Er, it was right. The image cache is also request URL keyed. https://html.spec.whatwg.org/multipage/images.html#the-list-of-available-images |
| 06:40 | <bradleymeck_> | I don't think that is really a decent comparison when it doesn't run code. Unclear on if your concern is about the HTTP requests being made or the Module Records being made |
| 06:43 | <bradleymeck_> | With ESM at least you have to do all the redirects before linking stuff anyway |
| 07:02 | <annevk> | bradleymeck_: found the last time this was discussed: https://github.com/whatwg/html/issues/613 |
| 07:15 | <JakeA> | annevk: (re Mozilla's CORB position) If we started the web again, no-cors wouldn't exist. Therefore I like CORB, and I'd like to see further efforts to limit no-cors. It has merits other than Spectre protection. |
| 07:15 | <JakeA> | (I'm not involved in the CORB effort internally at all, so that's a personal view) |
| 07:16 | <annevk> | JakeA: yeah, I'm definitely sympathetic to that point of view |
| 08:35 | <hsivonen> | Why does mimesniff allow whitespace bytes to precede "<?xml" when sniffing for XML? |
| 08:40 | <annevk> | hsivonen: perhaps a mistake, seems to have been there since it first got added |
| 08:40 | <annevk> | hsivonen: I'm also not sure how well implemented/tested mimesniff is in general |
| 08:40 | <hsivonen> | annevk: ok |
| 08:41 | <annevk> | hsivonen: the code in Firefox is really old, afaict |
| 08:41 | <annevk> | hsivonen: and this algorithm seems to mostly match it, so maybe it was written for Chrome/Safari |
| 09:47 | <Ms2ger> | Good to see people are still being helpful in bug trackers: https://bz.apache.org/bugzilla/show_bug.cgi?id=62273#c2 |
| 09:53 | <Ms2ger> | annevk, is it supposed to be possible to create a zero-sized ImageBitmap? |
| 09:55 | <annevk> | Ms2ger: dunno, need to get back to that PR at some point |
| 11:21 | <smaug____> | annevk: do you know if anyone is dealing with https://drafts.csswg.org/cssom related issues |
| 11:22 | <smaug____> | (context is https://bugzilla.mozilla.org/show_bug.cgi?id=1449584#c11) |
| 11:23 | <astearns_> | smaug____: we just appointed Emilio Cobos Γlvarez editor of CSSOM (an hour or so ago?) |
| 11:23 | <TabAtkins> | smaug____: We're dealing with them piecemeal at the moment; emilio just agreed to be the editor |
| 11:23 | <TabAtkins> | dangit, ninja'd |
| 11:23 | <astearns_> | :) |
| 11:26 | <smaug____> | nice |
| 11:36 | <ecobos> | smaug____: I guess now I'm to blame? :-) |
| 11:55 | <annevk> | ecobos: thanks for taking that on; hope you got a lot of spare time π |
| 12:39 | <nox> | https://github.com/w3c/css-houdini-drafts/issues/751 |
| 12:40 | <nox> | MFW a ticket becomes a wish list |
| 12:55 | <Newbie> | hi |
| 12:55 | <Newbie> | anybody |
| 12:56 | <Newbie> | trhy |
| 13:03 | <Ms2ger> | Bye |
| 13:22 | <annevk> | Domenic: counter example: https://github.com/mozilla/standards-positions/issues/82#issuecomment-380448688 π |
| 13:26 | <annevk> | Domenic: interested in reviewing https://github.com/w3c/web-platform-tests/pull/8449 too? |
| 13:26 | <Domenic> | annevk: I guess nobody else is stepping up, might as well. Assign me so I get an email I can flag? Will do today. |
| 13:29 | <annevk> | Thanks, I suspect most people moved on ten years ago or so |
| 13:29 | <annevk> | ajax() works just fine Anne, I don't know why you bother |
| 20:44 | <tobie> | foolip: ganesh move + rename seems to have been successful. |
| 20:44 | <tobie> | foolip: LMK if you see anything fishy |
| 22:14 | <foolip> | tobie: will do, thanks! |