| 05:28 | <hsivonen> | caitp: Thanks. It looks like the benchmark I did was not optimized away despite not using V8-specific black box. I guess neither V8 nor SpiderMonkey knows that TextDecoder and TextEncoder don't have side effects |
| 06:15 | <annevk> | crissov: prolly start drafting and maybe also figure out a testing strategy |
| 08:54 | <annevk> | crissov: maybe email Domenic and Hixie (email addresses should be at the bottom of the HTML Standard) if you need more advice |
| 08:57 | <annevk> | hsivonen: do you know if Fx copies the buffer passed to TextDecoder? We should maybe stop doing that if so |
| 09:06 | <yoav> | annevk: no rush, but would appreciate a review on https://github.com/whatwg/fetch/pull/549 when you have time :) |
| 09:06 | <annevk> | yoav: eta 4/5 weeks from now |
| 09:29 | <jochen__> | annevk: how would you feel about including firefox's concept of origin attributes in the HTML spec? |
| 09:30 | <jochen__> | annevk: as a generic point where other specs that want to modify SOP can hook in (such as suborigins or isolate-me)? |
| 10:42 | <annevk> | jochen__: not sure, if it makes sense architecturally to do it that way |
| 11:04 | <jochen__> | mind elaborating? |
| 11:20 | <annevk> | jochen__: it's not immediately clear to me origin attributes provide the right boundary for both |
| 11:23 | <annevk> | jochen__: I'd rather abstract after we figure out all the requirements |
| 11:24 | <jochen__> | well, clearly we'll never figure out all the requirements :) |
| 11:24 | <annevk> | jochen__: I'd also rather avoid a comparison like that in case there need to be differences |
| 11:24 | <jochen__> | so how do we go about figuring out a sufficient set of requirements |
| 11:25 | <annevk> | jochen__: I think by going through all the places that talk about origins and working through the implications |
| 11:26 | <annevk> | jochen__: no easy shortcuts |
| 11:26 | <jochen__> | that's basically the motivation for using origin attributes |
| 11:26 | <jochen__> | because they already did that |
| 11:28 | <annevk> | But origin attributes affect permissions |
| 11:28 | <annevk> | So not really the same |
| 11:28 | <annevk> | And they are not standardized, so I'm not sure what using them means |
| 11:29 | <annevk> | Either way we need to update all the standards |
| 12:29 | <zcorpan> | Domenic: annevk: Infra says "An ASCII whitespace is...", should that be s/An // ? |
| 12:36 | <ondras> | is there some spec and/or browser agreement for progressive jpegs vs. the "load" event? |
| 13:41 | <zcorpan> | MikeSmith: https://checker.html5.org/ gives me a 502 |
| 13:43 | <zcorpan> | ondras: see "If the resource type and data corresponds to a supported image format, as described below" in html |
| 13:46 | <ondras> | zcorpan: thanks |
| 13:49 | <ondras> | zcorpan: so apparently there is no way to detect the "the first lower-quality representation is available/visible" part of fetching the resource |
| 13:50 | <zcorpan> | ondras: correct |
| 13:51 | <zcorpan> | ondras: we could add progress events |
| 13:52 | <ondras> | no problem |
| 13:52 | <ondras> | I just wanted to understand the status quo. |
| 13:53 | <zcorpan> | it's a bit weird when there are 2 ongoing requests, but there's an idea to expose that in some way also |
| 13:54 | <zcorpan> | also see https://github.com/whatwg/html/pull/2332 |
| 13:54 | <MikeSmith> | zcorpan: thanks for the heads-up, https://checker.html5.org/ is back now |
| 13:55 | <zcorpan> | MikeSmith: 👌 |
| 15:55 | <zcorpan> | yoav: is the issue that as=fetch doesn't work yet, or would stop working with the change to ignore as=""? |
| 15:56 | <zcorpan> | (re blink-dev) |
| 16:12 | <yoav> | zcorpan: it doesn't work yet |
| 16:13 | <yoav> | the change would make it work instead of `as=""` |
| 16:58 | <zcorpan> | yoav: so maybe I'm still missing something. Why is the recommended replacement XHR/fetch(), rather than <link rel=preload as=fetch>? |
| 17:10 | <yoav> | zcorpan: because I totally missed it? :D |
| 17:10 | <yoav> | you're right, as=fetch is the answer |
| 17:11 | <zcorpan> | yoav: or I suppose the recommended replacement could be something like <link rel=preload as=fetch href=foo onerror="this.as = ''"> to make it work for both new and legacy |
| 17:12 | <zcorpan> | (or two <link>s if one doesn't want to rely on script at all) |
| 17:12 | <yoav> | yup, even though I'm not sure legacy is a big issue (but I'm sure you have more data than me on auto-updating rates) |
| 17:15 | <zcorpan> | yoav: I don't think it's a big issue, but, this would void the concern raised I believe |
| 17:15 | <yoav> | cool. |
| 17:16 | <yoav> | want to comment on the thread or should I? (while linking here and attributing you with the "new & legacy" solution) |
| 17:17 | <zcorpan> | please go ahead, I need to leave to the gym for a bit |
| 17:17 | <zcorpan> | no credit necessary ^_^ |
| 18:20 | <Domenic> | crissov: very exciting! Yeah, to second annevk, an initial draft and a testing strategy sounds like the right way to go. If it's shaping up into something that looks like a spec we can easily move it to WHATWG. Happy to help with the logistics of creating a spec document (using Bikeshed etc.) |
| 18:22 | <gsnedders> | Domenic: what's that in response to? |
| 18:22 | <Domenic> | 15:55:52 <crissov> Hi everyone! I'm still considering to volunteer as an editor for a PNG or APNG Living Standard at WhatWG |
| 18:23 | <Domenic> | 15:55:54 <crissov> https://lists.w3.org/Archives/Public/public-whatwg-archive/2016Dec/0027.html |
| 18:23 | <Domenic> | 15:56:18 <crissov> What exactly would I need to do to move this forward? |
| 18:23 | <gsnedders> | Domenic: thx |
| 18:23 | gsnedders | failed to find it scrolling up quickly |
| 18:30 | <yoav> | Does a living standard make sense for things like image formats? |
| 18:39 | <Domenic> | For sure, image formats get bug fixes all the time |
| 18:40 | <Domenic> | And PNG, it sounds like, gets new features even |
| 18:52 | <yoav> | Domenic: I know that at least for webp, support for "webp" mean very little due to old and busted androids |
| 18:53 | <yoav> | I guess that a living standard does make sense, but frozen snapshots serving as versions (and codified in MIME types) are also required |
| 18:54 | <yoav> | to make sure the ecosystem can survive changes (as feature detection is not an option, and browsers are not the only players) |
| 18:54 | <yoav> | Domenic: unrelated - RE https://github.com/whatwg/html/pull/2588 do you want me to add the required "potential destination" plumbing? |
| 19:10 | <Domenic> | yoav: yeah, either of us can do it; today is a holiday for me, and i have a cold so not sure i'll be recovered by tomorrow, but was planning on getting to it this week some time if you don't |
| 19:10 | <yoav> | ok, sounds good |
| 19:46 | <zcorpan> | yoav: for future reference, hit "Update" in the logs to get a date in the URL other than "today" :-) |
| 19:47 | <yoav> | oops! |