01:35:12 -BitBot- (wpt) [issue] karlcow opened #18947: [css-background] background shorthand and "/" slash syntax - https://git.io/Jeq14 01:52:59 -BitBot- (wpt) [PR] boazsender requested #18948 merge into master: Code of conduct suggestion - https://git.io/Jeq1r 02:37:32 -BitBot- (wpt) [PR] frivoal requested #18949 merge into master: [css-text] Adjust trailing space tests in response to https://github.com/w3c/csswg-drafts/issues/4180 - https://git.io/JeqMv 02:47:54 -BitBot- (wpt) [PR] boazsender requested #18950 merge into code-of-conduct.md: adds @jorydotcom and @boazsender as moderators to CoC - https://git.io/JeqMI 02:53:31 -BitBot- (wpt) [PR] chromium-wpt-export-bot requested #18951 merge into master: Initial Storage Access API IDL Changes - https://git.io/JeqML 02:56:49 -BitBot- (wpt) [PR] boazsender closed #18948: Code of conduct suggestion - https://git.io/Jeq1r 02:58:38 -BitBot- (wpt) [PR] marcoscaceres merged #18486 into master: Add tests for [[sharePromise]] internal slot - https://git.io/fjd2l 03:41:27 -BitBot- (wpt) [PR] moz-wptsync-bot merged #18944 into master: [Gecko Bug 1523582] Part 3 - Adjust GetLogicalSkipSides() to check non-fluid continuations. - https://git.io/Jeqo8 04:58:42 -BitBot- (wpt) [PR] chromium-wpt-export-bot merged #18920 into master: CSS WPT for computed font shorthand - https://git.io/JevxW 05:43:34 -BitBot- (wpt) [PR] heycam requested #18952 merge into master: Add test for invalid background shorthand position. - https://git.io/JeqyI 06:15:16 -BitBot- (wpt) [PR] autofoolip requested #18953 merge into master: Update interfaces/presentation-api.idl - https://git.io/JeqyP 06:15:20 -BitBot- (wpt) [PR] autofoolip requested #18954 merge into master: Update interfaces/push-api.idl - https://git.io/JeqyX 06:15:25 -BitBot- (wpt) [PR] autofoolip requested #18955 merge into master: Update interfaces/screen-capture.idl - https://git.io/Jeqy1 07:28:43 -BitBot- (wpt) [PR] foolip merged #18950 into code-of-conduct.md: adds @jorydotcom and @boazsender as moderators to CoC - https://git.io/JeqMI 07:34:47 -BitBot- (wpt) [PR] foolip merged #18946 into master: Update Device Orientation Event spec link - https://git.io/Jeqim 07:35:50 -BitBot- (wpt) [PR] foolip merged #18953 into master: Update interfaces/presentation-api.idl - https://git.io/JeqyP 07:36:27 -BitBot- (wpt) [PR] foolip merged #18954 into master: Update interfaces/push-api.idl - https://git.io/JeqyX 07:38:20 -BitBot- (wpt) [PR] foolip merged #18955 into master: Update interfaces/screen-capture.idl - https://git.io/Jeqy1 07:41:21 -BitBot- (wpt) [PR] foolip merged #18717 into master: Update interfaces/geolocation-API.idl - https://git.io/fjxv5 07:52:02 -BitBot- (rfcs) [PR] foolip reopened #17: RFC 17 - Code of Conduct - https://git.io/Jeq9q 08:57:04 -BitBot- (wpt) [PR] chromium-wpt-export-bot requested #18956 merge into master: [LayoutNG] Fix `vertical-align: top` and `bottom` - https://git.io/JeqHG 09:13:27 -BitBot- (wpt) [PR] foolip requested #18957 merge into master: Add test for rem/em dimensions based on :root font-size - https://git.io/JeqHa 09:14:32 jgraham: I have two GitHub Actions PRs, can you review? https://github.com/web-platform-tests/wpt/pull/18792 + https://github.com/web-platform-tests/wpt/pull/18924?https://github.com/web-platform-tests/wpt/pull/18924 09:14:44 https://github.com/web-platform-tests/wpt/pull/18924 09:17:08 foolip: Done, sorry for the delay 09:18:31 -BitBot- (wpt) [PR] foolip merged #18792 into master: [GitHub Actions] Pin the version of Ubuntu and checkout action - https://git.io/fjpK4 09:18:40 jgraham: not at all, gracias 09:19:05 -BitBot- (wpt) [PR] foolip merged #18924 into master: [GitHub Actions] trigger push workflows on push to master only - https://git.io/JeTGg 09:42:02 -BitBot- (wpt) [PR] rwlbuis requested #18958 merge into master: Fix Test - https://git.io/JeqQk 09:45:23 -BitBot- (wpt) [PR] chromium-wpt-export-bot requested #18959 merge into master: CSS: Test computed values for CSS Motion path properties - https://git.io/JeqQY 09:58:28 -BitBot- (wpt) [PR] rwlbuis merged #18958 into master: Fix Test - https://git.io/JeqQk 10:02:14 -BitBot- (wpt) [PR] fred-wang requested #18960 merge into master: MathML: test dynamic changes for on[event] attributes. - https://git.io/JeqQB 10:21:04 -BitBot- (wpt) [PR] fred-wang merged #18960 into master: MathML: test dynamic changes for on[event] attributes. - https://git.io/JeqQB 10:31:46 -BitBot- (wpt) [PR] chromium-wpt-export-bot requested #18961 merge into master: [Trusted Types] Test current behavior of eval(TT) in WPTs. - https://git.io/Jeq7t 11:07:50 -BitBot- (wpt) [PR] chromium-wpt-export-bot merged #18915 into master: CSS: counter-reset/increment inherit clears old values - https://git.io/fjjSP 11:12:56 foolip: ping 11:15:43 dotnetCarpenter: Hi I'm here, but I'm heading to lunch 11:16:10 jgraham: https://tools.taskcluster.net/groups/bcXnBJKTQ1eUo9GX18m14w might be an emergency, looks like widespread failures due to the event data :{ 11:29:24 foolip: ahh - I just saw that you posted in the mailing list.. That answer my question :) Is there anything I need to fix in my PR or should I just be patient? 11:34:27 great news that Safari is passing https://github.com/web-platform-tests/wpt/pull/18941 then at least 2 bugs can be closed. https://bugs.webkit.org/show_bug.cgi?id=117680 and https://bugs.webkit.org/show_bug.cgi?id=123012 11:34:47 -BitBot- (wpt) [PR] dotnetCarpenter closed #18941: [css-values] Reftest for percentage root font size and rem units - https://git.io/JeqVO 11:35:28 -BitBot- (wpt) [PR] dotnetCarpenter reopened #18941: [css-values] Reftest for percentage root font size and rem units - https://git.io/JeqVO 11:58:12 foolip: Thank you so much for the review! I got a lot to say but I have to be AFK for an hour or two. I will write on github 12:01:23 dotnetCa1penter: hopefully Safari passing isn't spurious, let's see when the test is refined a bit :) 12:03:36 foolip: BTW, please use devtools to measure the difference. The difference is 187.5px * 187.5px in Chrome but should be 100px * 100px in every other browser 12:05:01 dotnetCa1penter: if it result should be 100x100, maybe the mismatch ref isn't needed? 12:08:56 foolip: Bleh, this is the thing where PRs are using the .taskcluster.yml from their HEAD and the python files from master 12:09:22 jgraham: oh wonderful... 12:10:30 jgraham: so for a change to .taskcluster.yml to be robust, it has to work with an arbitrarily old state of the rest of the tree? seems tricky :) 12:11:37 I filed a TC bug about it at some point, but I don't think there were any takers :) 12:11:41 foolip: I wanted to reference https://web-platform-tests.org/ docs, but I really have to go and meet a client. The ultra short version is that the docs says that re-using existing refs is better, so I changed the PR to use an existing 100*100 and changed my ref to be a mismatch highlighting what actually happens in Chrome (which is wrong according to spec). I have some comments about the wording 12:11:47 in the specs as well.. I can see why Chromium developers would implement this behavior.. section about minimal relative font-size. I also thought that w3 Candidate Recommendation was more appropiate as spec ref than w3 Editor's Draft - but I probably misunderstood the w3 conventions here 12:13:57 -BitBot- (wpt) [PR] jgraham requested #18962 merge into master: Use TASK_EVENT from the environment when present - https://git.io/JeqFo 12:14:04 foolip: https://github.com/web-platform-tests/wpt/pull/18962 12:14:49 dotnetCa1penter: the downside of a mismatch ref, as you can probably imagine, is that it's very easy to make something different for the wrong reason. now you're not using just a mismatch, but it sounds like you're using it to illustrate what the actual outcome in Chrome is. IMHO that's not necessary, the actual results will be available as a screenshot on failure. 12:15:27 foolip: true 12:15:29 jgraham: approved, just as well to merge right away? 12:16:00 * dotnetCa1penter have to go and don't have a server to listen to IRC, so all messages while I'm not here will be void 12:19:19 -BitBot- (wpt) [PR] chromium-wpt-export-bot requested #18963 merge into master: CSS: Serialize border-radius with / - https://git.io/JeqFP 12:26:06 -BitBot- (wpt) [PR] jgraham merged #18962 into master: Use TASK_EVENT from the environment when present - https://git.io/JeqFo 12:26:12 foolip: Yeah I merged it 12:26:55 -BitBot- (wpt) [issue] jgraham closed #18947: [css-background] background shorthand and "/" slash syntax - https://git.io/Jeq14 12:26:55 -BitBot- (wpt) [PR] jgraham merged #18952 into master: Add test for invalid background shorthand position. - https://git.io/JeqyI 12:31:34 -BitBot- (wpt) [PR] saschanaz requested #18964 merge into master: [web-share] Use HTTPS URLs if possible - https://git.io/JeqFb 12:32:00 foolip: If there are more PRs affected, close and reopen should hopefully fix 12:36:46 -BitBot- (wpt) [PR] chromium-wpt-export-bot merged #18959 into master: CSS: Test computed values for CSS Motion path properties - https://git.io/JeqQY 13:20:24 hah, https://github.com/web-platform-tests/wpt/pull/14604 got rebase-merged so we have all the fixup! fixup! fixup! fixup! squash! fun on master 13:23:39 jgraham: ^ 13:24:03 tl;dr: we should be at least somewhat careful when we rebase-merge PRs onto master 13:24:30 I've sort of given up on a pretty history. Even if only the squash option is allowed, most people still just accept the default message with a bullet point per commit, which ends up making little sense usually. 13:24:55 I guess I'd prefer the default to be the PR title/description. 13:25:15 Let's built a commit bot to fix this pressing problem! 13:25:16 The squash implementation is "we don't give a flying flamingo" levels of bad 13:26:01 jgraham: Other than the commit message, is anything else wrong? It usually takes quite little editing of it (like deleting the body) for it to be nice 13:26:10 gsnedders: YEah, that was a mistake, but one that's easy to make 13:27:18 foolip: Well mostly that the commit message just never makes sense, and ends up being rewritten by the committer rather than the patch author, although I don't think that's actually recorded anywhere (maybe one is supposed to assume that the committer is responsible for the message) 13:27:45 It doesn't even ignore "fixup!" commits when making the proposed message 13:27:52 jgraham: the only way the system can work is if the person who send the PRs rewrites the history, yes 13:28:08 Which tells me they spent as little time on the feature as was required to release it and haven't touched it since 13:30:59 (if we actually had a bot I might propose that every PR would land as a ff merge with a merge commit, so that the history would accurately record the history of PRs landing, without needing the "tag every merge" hack that we're currently using) 13:32:40 (note that the proposal doesn't cause much history complexity, because every commit that lands has one parent apart from the guaranteed-empty merge commits, which have exactly two; the previous merge commit and the head of the PR that landed. And the PR commits are always based on the previous PR that landed) 13:33:00 yeah if we could add some simple automation there I'd be all for that 13:33:07 (anyway, not going to happen, but the only "sensible" way to organise git history) 13:45:55 -BitBot- (wpt) [PR] chromium-wpt-export-bot requested #18965 merge into master: Fix background-position-interpolation.html for initial/unset - https://git.io/JeqpU 14:01:42 A commit bot would be sort of fun, but would need some other strong selling points in addition to nicer history. Maybe guaranteeing that stuff doesn't start failing on master :) 14:01:44 -BitBot- (wpt) [PR] chromium-wpt-export-bot merged #18961 into master: [Trusted Types] Test current behavior of eval(TT) in WPTs. - https://git.io/Jeq7t 14:02:26 (Except that requires pinning browser versions, which we can't do without hosting our own Chrome packages somewhere.) 14:03:29 I don't know how high our commit volume is compared to what's required for the "not rocket science" approach to CI 14:04:02 c.f. https://graydon.livejournal.com/186550.html 14:06:49 jgraham: is the gist of it to test the exact candidate master, with the effect of having to test all changes sequentially instead of in parallel? 14:07:18 foolip: Yeah 14:08:28 jgraham: dunno, but my hunch is that running the tests we actually trigger now, it's not a very high volume of the slowest tests, which I think are ~15min for some tools integration tests? 14:08:53 But then a heuristic for what tests to run can't really achieve a guaranteed green master. 14:09:10 But... the commit bot could do more work than is done for PRs. 14:09:30 Let's stop talking before we start thinking a commit bot our highest priority. :P 14:09:32 There's a tension with latency (obviously) because you can only land one change per ci cycle, and with workload because to get early feedback to authors you probably also want to run the tests based on the PR head 14:09:49 Yeah, I don't think this is the highest priority 14:09:50 ) 14:09:52 :) 14:10:07 what, this definitely sounds like a P0 14:10:20 It's clearly lower priority than a proper decision task for tc, which we've also put off for some time 14:22:32 In terms of CI, I think useful feedback based on the test outcomes is the most valuable thing we could add at this point. The sort of thing lukebjerring has been building but we don't have a clear plan for how to balance the noisiness vs. usefulness yet. On TPAC agenda. 14:25:57 Yeah I agree that's useful. GitHub checks has been underwhelming in terms of its ability to surface this kind of thing 14:26:46 The CI integrations are all still lacklustre and I don't think that many people look at the results unless there's an actual error 15:27:53 -BitBot- (wpt) [issue] gsnedders opened #18966: Failing to restart browser doesn't fail job - https://git.io/JemJT 15:28:32 -BitBot- (wpt) [PR] gsnedders requested #18967 merge into master: Fix #18966: Upgrade various ERRORs to CRITICAL - https://git.io/JemJI 15:48:27 -BitBot- (wpt) [PR] chromium-wpt-export-bot merged #18907 into master: Move forced colors tests to wpt - https://git.io/fjjPm 15:58:57 -BitBot- (wpt) [PR] chromium-wpt-export-bot requested #18968 merge into master: Add feature-detectable API for Scroll To Text - https://git.io/JemUO 16:00:37 -BitBot- (wpt) [PR] jgraham requested #18969 merge into master: Revert "Add async dispatch event flag to Chromedriver perform action" - https://git.io/JemUC 16:11:31 -BitBot- (wpt) [issue] gsnedders closed #18966: Failing to restart browser doesn't fail job - https://git.io/JemJT 16:11:32 -BitBot- (wpt) [PR] gsnedders merged #18967 into master: Fix #18966: Upgrade various ERRORs to CRITICAL - https://git.io/JemJI 16:15:57 jgraham: I think the only near term solution for this would be to have one custom check integrating info from the others and supporting our custom rules for what to flag. Long term the only option I could see is running everything on a single CI systems that also supports summary tasks and fancy integration with checks output. 16:17:16 I dislike that non-critical jobs make the PR show as [red cross]. 16:18:01 Azure Pipelines and GitHub Actions both have the OS support and job dependency stuff, but nothing at all has good checks integration. 16:18:07 gsnedders: yeah... 16:27:37 foolip: I'm back. I'm going to implement the changes you requested on github and push 16:30:33 dotnetCarpenter: uh, let's fix the docs! :) 16:35:19 dotnetCarpenter: I think my recommendation would be write a new test using testharness.js (see https://web-platform-tests.org/writing-tests/testharness.html) based on Trevor Burnham's test? 16:35:36 gsnedders: Perhaps I just took it too literally. It is under a headline saying "More practice" 16:35:54 dotnetCarpenter: we should clarify it, whatever! if it confused you, it'll confuse others. 16:36:54 gsnedders: thanks for the info about testharness.js, I will look into it after refactoring the current reftest 16:56:21 dotnetCarpenter: I'm off work now, but might take a look tonight if you update the PR 16:58:16 I dont' understand the result https://wpt.fyi/results/css/css-values/percentage-rem-low.html?label=pr_head&max-count=1&pr=18941 of my latest commit to https://github.com/web-platform-tests/wpt/pull/18941 16:58:27 why does it now fail in firefox? 16:58:55 dotnetCarpenter: have you tried looking at the screenshot comparison? 17:00:29 foolip: thanks. No stress please. I will also be here tomorrow 17:00:40 gsnedders: yes 17:00:59 dotnetCarpenter: what do you see as the difference there? 17:01:52 dotnetCarpenter: (turning off "Highlight diff" and looking at the actual and expected screenshots makes this clearer) 17:02:27 gsnedders: Well it seems like a margin issue - https://wpt.fyi/analyzer?screenshot=sha1%3A26802d00e57a78123a7a789c81502f0cafee27ab&screenshot=sha1%3A2a2c7612408301c9eee369dbe1ebeb2cbc91d8c0 17:02:41 not related to the test... 17:03:13 dotnetCarpenter: note there's text on the expected screenshot but not on the actual screenshot 17:03:21 ahh wait 17:03:33 yes - I see that now. 17:08:05 dotnetCarpenter: it will have passed before because the requirement was the match *or* mismatch passed… and it's very easy for a mismatch to pass! 17:08:33 -BitBot- (wpt) [PR] jgraham merged #18969 into master: Revert "Add async dispatch event flag to Chromedriver perform action" - https://git.io/JemUC 17:11:52 foolip: But even then I'm not sure people click through to look at the check output unless it's a blocking error 17:12:17 For things like "this test causes an error in some browsers" it's unlikley that will end up being a blocking error 17:15:30 foolip: https://dev.azure.com/web-platform-tests/wpt/_build/results?buildId=30398 is now failing 17:38:13 jgraham: The large gap between "all passing and consistent cross-browser" and "definitely a test authoring problem" is the challenge as I see it here. Perhaps the most useful feedback will have to be on an FYI basis, in which case we'll have to design around that... but how? Having checks that pass with warnings would be useful, but that's a GitHub UI feature request. 17:38:45 gsnedders: Sweet! How I've missed that red :) 17:57:54 FYI, my latest commit results shows that the test pass in firefox, as expected. And fails in Chrome, as expected. In Safari the test fails but not only is the square too big (as in Chrome) but the ref text, which is styled with font-size:initial;, is also too small. The wrong text font-size is not really part of the test. 17:57:59 https://wpt.fyi/results/css/css-values/percentage-rem-low.html?label=pr_head&max-count=1&pr=18941 17:59:30 Should I change the ref text styling to font-size: 16px;, so that does not impact if the test fails or succeed? I'm not sure if there is a test for font-size: initial; but there probably should be since Sadari differs from Chrome and Firefox. 17:59:55 ~Sadari~ Safari 18:10:03 Right, I'm off for now but will check back with you tomorrow. Thank you for all the feedback. It has been a very nice experience indeed. 18:12:05 -BitBot- (wpt) [PR] jugglinmike requested #18970 merge into master: [docs] Simplify README - https://git.io/JemID 18:14:09 I take it back about Safari - perhaps I was seeing a cahed version of the screenshot - it look the exact same as in chrome now (still not if I visit the same page in firefox, hence I believe it to about caching) 18:35:35 foolip: Right, I don't know what the solution is if we're still forbidden from adding comments 18:39:05 -BitBot- (wpt) [PR] fred-wang requested #18971 merge into master: Add more MathML tests for tabIndex - https://git.io/JemLc 18:41:45 jgraham: if we reduce the number of checks visible on every PR it's possible a neutral outcome check would stand out enough, but if not I don't think we should exclude comments as part of the solution. Or perhaps review comments, that is comments on a specific file, which can be resolved. We must find a way :D 18:55:09 foolip: Well I'm not in the anti-comment camp to begin with :) 18:55:50 I don't think that review comments are less annoying; they still generate a notification 18:55:51 -BitBot- (wpt) [PR] LukeZielinski requested #18972 merge into master: Change documentation to break the assumption that Ahem is a system font. - https://git.io/JemLM 18:56:57 But having said that, I wonder how it would work out to have a bot that set "changes required" if a check fails so that the review would have to be explictly dismissed to indicate that you saw the issue but chose to ignore it 18:57:17 It would only be worthwhile for things that are high probability errors ofc 19:15:51 -BitBot- (wpt) [PR] chromium-wpt-export-bot merged #18942 into master: [LayoutNG] Fix line breaking behavior after out-of-flow objects - https://git.io/Jeqr1 19:19:04 jgraham: That's a novel idea, I like it. Makes it easier to write the instructions for how to proceed in the bot's review comment. Other than that it's similar to a check with an action to dismiss as lukebjerring built, which I liked very much and perhaps could achieve a similar effect if the check output was very clear about what to do next. 19:26:01 -BitBot- (wpt) [PR] chromium-wpt-export-bot requested #18973 merge into master: Make WebXR WPT tests use viewOffset when appropriate - https://git.io/JemtW 19:26:39 -BitBot- (wpt) [PR] chromium-wpt-export-bot requested #18974 merge into master: [LayoutNG] Fix `vertical-align: top/bottom` with leading - https://git.io/Jemtl 19:32:17 -BitBot- (wpt) [PR] Hexcles merged #18968 into master: Add feature-detectable API for Scroll To Text - https://git.io/JemUO 19:36:26 -BitBot- (wpt) [PR] LukeZielinski requested #18975 merge into master: [wip] See what happens if we make checkLayout wait for fonts. - https://git.io/Jemti 19:53:48 -BitBot- (wpt) [PR] chromium-wpt-export-bot merged #18965 into master: Fix background-position-interpolation.html for initial/unset - https://git.io/JeqpU 20:15:47 -BitBot- (wpt) [PR] chromium-wpt-export-bot requested #18976 merge into master: TransferToImageBitmap should reset the bitmap to transparent black - https://git.io/JemqK 20:42:15 -BitBot- (wpt) [PR] chromium-wpt-export-bot merged #18963 into master: CSS: Serialize border-radius with / - https://git.io/JeqFP 20:51:27 -BitBot- (wpt) [PR] moz-wptsync-bot requested #18977 merge into master: [Gecko Bug 1568171] Fix handling of load events if document.open() is called while async scripts are still loading for the original pageload. - https://git.io/JemmR 21:06:06 -BitBot- (wpt) [PR] chromium-wpt-export-bot merged #18973 into master: Make WebXR WPT tests use viewOffset when appropriate - https://git.io/JemtW 21:54:56 -BitBot- (wpt) [PR] chromium-wpt-export-bot requested #18978 merge into master: code health: mark manual WPT tests in css-transitions - https://git.io/JemYw 22:10:24 -BitBot- (wpt) [PR] emilio merged #18866 into master: [css-text-decor-4] Fix the expected serialization of text-decoration. - https://git.io/fjjet 22:50:03 -BitBot- (wpt) [PR] chromium-wpt-export-bot requested #18979 merge into master: [css-grid] Additional cases where the override width must be cleared - https://git.io/JemOu