| 03:00 | <devsnek> | is there some secret way to tell browsers that when they resize they should try to keep the content that was being viewed on screen |
| 03:47 | <domfarolino> | I wish |
| 09:05 | <JakeA> | Mozilla folks: Is there some way to put iframes in a different thread/process to the main page? |
| 09:05 | <JakeA> | (in Firefox) |
| 09:06 | <jgraham> | JakeA: Enable fission |
| 09:07 | <jgraham> | https://wiki.mozilla.org/Project_Fission |
| 09:08 | <jgraham> | (or the COOP/COEP stuff I suppose) |
| 10:16 | <JakeA> | cheers! |
| 11:03 | <MikeSmith> | gsnedders: jgraham: annevk: in the html5lib test suite, do you recall why there is a need for some tests to have the doubleEscaped thing, where code points are written with, e.g., "\\u0000" rather than just "\u0000"? |
| 11:15 | <gsnedders> | MikeSmith: IIRC some JSON parsers not coping with null, plus also the whole "does JSON allow lone surrogates" fun |
| 11:37 | <JakeA> | annevk: Serious question: Could we just spec `history.length` as a blocking cross-process getter, like `localStorage`, and write it off as 'legacy'? |
| 12:06 | <MikeSmith> | gsnedders: ah OK, thanks |
| 12:08 | <MikeSmith> | gsnedders: so if I were to pre-process the test files to globally replace "\\uNNNN" with "\uNNNN", I wouldn’t be breaking any of the tests? |
| 12:08 | <MikeSmith> | I mean, if I did that, the tests would still be testing the same thing they are intended to test? |
| 12:18 | <gsnedders> | MikeSmith: p sure; I think in theory "\\\uNNNN" should have different behaviour, but I doubt anyone does that |
| 12:18 | <MikeSmith> | yeah |
| 12:18 | <MikeSmith> | thanks! |
| 12:30 | <zcorpan> | Hello whatwg |
| 12:36 | <gsnedders> | hello zcorpan |
| 12:43 | <MikeSmith> | heya zcorpan |
| 12:43 | <zcorpan> | Has something interesting happened in the past 4 weeks? :) |
| 12:45 | <gsnedders> | I started a new job? :) |
| 12:51 | <zcorpan> | ah yes gsnedders! Congrats \o/ |
| 17:55 | <annevk> | JakeA: maybe, doesn’t sound like what we want long term, but reasonable given it’s the status quo |
| 18:00 | <Domenic> | You could probably define it as sync cross-agent-cluster-blocking but also approximate. E.g. "An implementation-defined approximation of the value computed as follows: ... NOTE: this is implementation-defined because in multi-process browsers, the above algorithm requires synchronous inter-process communication, which isn't always feasible." |
| 20:39 | <annevk> | I would prefer trying to tie it to the top-level origin/site, fwiw |
| 21:19 | <gsnedders> | JakeA: what's most usage of history.length like? how much is just checking if it's non-empty? |
| 21:20 | <gsnedders> | also how much cross-origin (v. the top-level) usage is there, which AFAIK is only as far as current plans for multi-process go? how much would break if it lied? |