Summary: | Move message-port-multi.js from resources to script-tests | ||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Product: | WebKit | Reporter: | Shinichiro Hamaji <hamaji> | ||||||||||
Component: | Tools / Tests | Assignee: | Nobody <webkit-unassigned> | ||||||||||
Status: | RESOLVED FIXED | ||||||||||||
Severity: | Normal | CC: | atwilson, commit-queue, eric | ||||||||||
Priority: | P2 | ||||||||||||
Version: | 528+ (Nightly build) | ||||||||||||
Hardware: | PC | ||||||||||||
OS: | OS X 10.5 | ||||||||||||
Attachments: |
|
Description
Shinichiro Hamaji
2009-09-28 00:42:11 PDT
Created attachment 40220 [details]
Patch v1
Comment on attachment 40220 [details] Patch v1 Clearing flags on attachment: 40220 Committed r48823: <http://trac.webkit.org/changeset/48823> All reviewed patches have been landed. Closing bug. This test contains asynchronous tests, so including js-test-post.js is incorrect. Note that if you look at the test expections, it inserts "TEST COMPLETE" in the middle of the test output, which is not correct. (In reply to comment #4) > This test contains asynchronous tests, so including js-test-post.js is > incorrect. > > Note that if you look at the test expections, it inserts "TEST COMPLETE" in the > middle of the test output, which is not correct. I see. So, we can just revert my patch? Yes, it's a known issue that the js testing framework stuff doesn't handle async tests well. I thought I filed a bug about this a while back. Basically we'd like to eventually add an "imDone()" method of some sort that would print the TEST COMPLETED stuff for you. Many tests have already "hacked" this by removing js-test-post.js and done their own TEST COMPLETED writing. Eventually we should just make this function part of js-test-pre.js and have js-test-post call it when the test is a waitUntilDone() test. It sounds like you were trying to get this test to be generated from TEMPLATE.html, but it's probably not appropriate for this test since it doesn't follow the standard form. I don't actually understand what the motivation behind the test was - if it was just an attempt at cleanup, then yeah, I'd revert it. Created attachment 40309 [details]
Revert patch v1
I think we're trying to slowly improve the "standard form" over time. The original dream of all the .js tests was that we wouldn't check in the .html files at all. I'm not sure we'll ever reach that dream though. Definitely js-test-pre.js and js-test-post.js need losts of love to support more types of testing (like async tests, XHTML tests, SVG tests, etc.) Created attachment 40342 [details]
Revert patch v2
(In reply to comment #10) > Created an attachment (id=40342) [details] > Revert patch v2 I've updated the revert patch so it has a comment to mention we don't use js-test-post. Sorry for confusion by this bug. Comment on attachment 40342 [details]
Revert patch v2
Rejecting patch 40342 from commit-queue.
Failed to run "['WebKitTools/Scripts/run-webkit-tests', '--no-launch-safari', '--quiet', '--exit-after-n-failures=1']" exit_code: 1
Running build-dumprendertree
Running tests from /Users/eseidel/Projects/CommitQueue/LayoutTests
Testing 11355 test cases.
fast/events/message-port-multi.html -> failed
Exiting early after 1 failures. 6051 tests run.
110.37s total testing time
6050 test cases (99%) succeeded
1 test case (<1%) had incorrect layout
2 test cases (<1%) had stderr output
Created attachment 40346 [details]
Revert patch v3
(In reply to comment #13) > Created an attachment (id=40346) [details] > Revert patch v3 Sorry, I used patch -R and resources/*.js wasn't tracked by git. I created the patch again using git mv. Comment on attachment 40346 [details] Revert patch v3 Clearing flags on attachment: 40346 Committed r48926: <http://trac.webkit.org/changeset/48926> All reviewed patches have been landed. Closing bug. |