| 02:27 | <jermw> | Hi everyone. How can I keep track of the diffs in the HTML Living Standard? Let's say I want to: see what new features exist from the last time I reviewed it; see what is on the roadmap for inclusion; see what has changed/dropped etc. How would I do that? |
| 04:43 | <MikeSmith> | jermw: look through the commit log for the repo, and look through the issue tracker |
| 04:56 | <MikeSmith> | Domenic: thanks ーfiled https://github.com/WICG/feature-policy/issues/245 |
| 05:31 | <rniwa> | jermw: I think there's a way to see the diff between two git hashes |
| 05:31 | <rniwa> | I forgot how to do that though |
| 05:31 | <rniwa> | maybe MikeSmith would know |
| 05:35 | <MikeSmith> | https://blog.github.com/2010-03-01-introducing-github-compare-view/#how-to-customize-the-commit-range |
| 07:36 | <annevk> | There is a rendered diff for each PR, but you’d have to locate that |
| 10:04 | <jermw> | ok thx everyone |
| 12:20 | <hsivonen> | I'm looking for a linkable reliable source that documents that V8 has the optimization of not storing the high half of every UTF-16 code unit if the string is entirely in the Latin1 range. Where should I look? |
| 16:45 | <annevk> | hsivonen: https://chromium.googlesource.com/v8/v8/+/master/src/objects/string-inl.h#152 |
| 16:45 | <annevk> | hsivonen: (though maybe not link to master) |
| 16:45 | <annevk> | hsivonen: not the greatest source though, I think there was a post somewhere at some point that explained it, but can't find it |
| 17:16 | <hsivonen> | annevk: yeah, I located source that checks for one-byte vs. two-byte but I failed to locate a code comment that says outright what one-byte is |
| 17:18 | <annevk> | hsivonen: I did find a SpiderMonkey post which says one of the reasons for not doing UTF-8 is a regexp engine copied from V8 that already handles Latin1… Lost the link, but from 2014 by Jan de Mooi |
| 17:33 | <hsivonen> | annevk: Thanks. Not trivially googlable. I think I'll just assert things about V8 without a link. |
| 17:34 | <hsivonen> | annevk: I think I know why there are still two failing UTF-32 text/plain test cases on WPT. I'm guessing this hack is to blame: https://bugzilla.mozilla.org/show_bug.cgi?id=631751 |
| 17:34 | <annevk> | hsivonen: hmm they were all passing for me iirc |
| 17:35 | <annevk> | hsivonen: or outside encoding/? |
| 17:35 | <hsivonen> | annevk: https://wpt.fyi/results/encoding/unsupported-encodings.any.html?label=experimental |
| 17:35 | <hsivonen> | hmm. the worker variant fails too and the HTML parser isn't run on Workers |
| 17:36 | <hsivonen> | so maybe my guess is wrong |
| 17:37 | <annevk> | How does Chrome suddenly pass? |
| 17:38 | <hsivonen> | no, my guess is wrong |
| 17:38 | <hsivonen> | the test is wrong, too |
| 17:38 | <hsivonen> | the test expects text/plain in XHR to default to windows-1252, but it defaults to UTF-8 |
| 17:40 | <annevk> | hsivonen: oh this test was not touched I think in that PR |
| 17:40 | <annevk> | hsivonen: I can fix this tomorrow if you file something somewhere to remind me |
| 17:40 | <hsivonen> | jgraham: thanks for the wpt github invitation. |
| 17:44 | <hsivonen> | Filed as https://github.com/web-platform-tests/wpt/issues/14082 |
| 17:45 | <annevk> | Great! |
| 19:10 | <jorendorff> | Domenic: I have a dumb question about Streams. Is it really necessary to pass forAuthorCode everywhere, or is it a function of what kind of reader exists? |
| 19:10 | <jorendorff> | i'm wondering if i can implement it as a bit per stream |
| 19:44 | <jorendorff> | filed #964 |