Bug 216301 - Import selection web platform tests
Summary: Import selection web platform tests
Status: RESOLVED FIXED
Alias: None
Product: WebKit
Classification: Unclassified
Component: New Bugs (show other bugs)
Version: WebKit Nightly Build
Hardware: Unspecified Unspecified
: P2 Normal
Assignee: Alex Christensen
URL:
Keywords: InRadar
Depends on:
Blocks:
 
Reported: 2020-09-08 19:00 PDT by Alex Christensen
Modified: 2020-09-14 12:45 PDT (History)
7 users (show)

See Also:


Attachments
Patch (6.85 MB, patch)
2020-09-08 19:03 PDT, Alex Christensen
youennf: review+
Details | Formatted Diff | Diff

Note You need to log in before you can comment on or make changes to this bug.
Description Alex Christensen 2020-09-08 19:00:26 PDT
Import selection web platform tests
Comment 1 Alex Christensen 2020-09-08 19:03:46 PDT
Created attachment 408300 [details]
Patch
Comment 2 EWS Watchlist 2020-09-08 19:05:12 PDT
This patch modifies the imported WPT tests. Please ensure that any changes on the tests (not coming from a WPT import) are exported to WPT. Please see https://trac.webkit.org/wiki/WPTExportProcess
Comment 3 Alexey Proskuryakov 2020-09-08 20:04:46 PDT
Comment on attachment 408300 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=408300&action=review

> LayoutTests/ChangeLog:3


Why?
Comment 4 Alexey Proskuryakov 2020-09-08 20:05:33 PDT
I know what the cost is - slower tests and thus lower engineering efficiency. What is the benefit?
Comment 5 Alex Christensen 2020-09-08 22:18:11 PDT
My thought was that this could be a good first step in fixing the failures.  Alternatively, I could just keep this on a local branch and make the fixes that fix these failures with a smaller number of tests I write myself.

Would skipping these tests in debug builds or in wk1 make you less hesitant to add these?
Comment 6 youenn fablet 2020-09-08 23:03:08 PDT
Comment on attachment 408300 [details]
Patch

Skipping timing out tests is always preferable than running them.
It is also better to import all tests in a folder: we can fix them progressively and we do not have to spend the time to rediscover they are timing out on each reimport.

Looking at some of these tests, they end up with more than 5000 subtests...
Alex, please verify these tests pass in debug.
Comment 7 Alex Christensen 2020-09-09 09:15:09 PDT
Added iOS expectations for selection-select-all-move-input-crash.html too.
http://trac.webkit.org/r266784
Comment 8 Radar WebKit Bug Importer 2020-09-09 09:16:20 PDT
<rdar://problem/68576888>
Comment 9 Alex Christensen 2020-09-09 09:27:05 PDT
http://trac.webkit.org/r266785
Comment 10 Alexey Proskuryakov 2020-09-09 09:52:49 PDT
(In reply to Alex Christensen from comment #5)
> My thought was that this could be a good first step in fixing the failures. 

Makes sense, and we've been taking this approach a lot. I don't think that we have a great way to make these decisions, but "more tests is better" feels wrong to me at this point.

> Alternatively, I could just keep this on a local branch and make the fixes
> that fix these failures with a smaller number of tests I write myself.
> 
> Would skipping these tests in debug builds or in wk1 make you less hesitant
> to add these?

Since this is all in WebCore, skipping in WK1 seems like a good idea. But I don't want to slow down WK2 tests either, and certainly we want to catch debug assertions too.