Bug 98949 - IndexedDB: setTimeout in layout tests causes flakiness in multiprocess ports
Summary: IndexedDB: setTimeout in layout tests causes flakiness in multiprocess ports
Status: RESOLVED FIXED
Alias: None
Product: WebKit
Classification: Unclassified
Component: New Bugs (show other bugs)
Version: 528+ (Nightly build)
Hardware: Unspecified Unspecified
: P2 Normal
Assignee: Joshua Bell
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2012-10-10 13:10 PDT by Joshua Bell
Modified: 2012-10-16 09:52 PDT (History)
4 users (show)

See Also:

Attachments
Patch (11.62 KB, patch)
2012-10-10 13:12 PDT, Joshua Bell 		no flags Details | Formatted Diff | Diff
Patch (17.66 KB, patch)
2012-10-15 16:34 PDT, Joshua Bell 		no flags Details | Formatted Diff | Diff
Patch (20.96 KB, patch)
2012-10-15 17:24 PDT, Joshua Bell 		no flags Details | Formatted Diff | Diff
Show Obsolete (2) View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Joshua Bell 2012-10-10 13:10:19 PDT 
IndexedDB: setTimeout in layout tests causes flakiness in multiprocess ports
Comment 1 Joshua Bell 2012-10-10 13:12:59 PDT 
Created attachment 168059 [details]
Patch
Comment 2 Joshua Bell 2012-10-10 13:15:14 PDT 
NOTE: WIP - intversion-close-between-events.html hasn't been fixed yet. It's unclear if that test is really testing anything distinct from intversion-close-in-oncomplete.html and intversion-close-in-upgradeneeded.html other than the race condition.
Comment 3 Joshua Bell 2012-10-15 16:34:47 PDT 
Created attachment 168806 [details]
Patch
Comment 4 Joshua Bell 2012-10-15 16:35:21 PDT 
dgrogan@ - please take a look?
Comment 5 David Grogan 2012-10-15 16:59:06 PDT 
Comment on attachment 168806 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=168806&action=review

LGTM

Thanks for fixing these

> LayoutTests/storage/indexeddb/resources/intversion-close-between-events.js:66
> +    if (!didCallCloseDB) {

We should probably test the other order also, if didCallCloseDB then evalAndExpectException("db.transaction(...)").
Comment 6 Joshua Bell 2012-10-15 17:24:09 PDT 
Created attachment 168812 [details]
Patch
Comment 7 Joshua Bell 2012-10-15 17:34:27 PDT 
Comment on attachment 168812 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=168812&action=review

> LayoutTests/fast/js/resources/js-test-pre.js:148
> +function evalAndLog(_a, _quiet)

The _quiet is so that if the text being eval'd is something like "quiet += 1" it doesn't collide.

(Maybe we should change these functions to use an indirect eval?)

> LayoutTests/fast/js/resources/js-test-pre.js:-530
> -                

This was actually a line full of space characters. I just deleted the line rather than fight with my editor that likes to delete trailing whitespace.

> LayoutTests/storage/indexeddb/mozilla/resources/create-index-unique.js:-41
> -    index = evalAndLog("index = objectStore.createIndex(indexName, indexKeyPath, { unique: true });", "IDBDatabaseException.CONSTRAINT_ERR");

Bogus second argument now being treated as |quiet|. I couldn't find any other examples in layout tests, but may have missed some.
Comment 8 Joshua Bell 2012-10-15 17:34:49 PDT 
tony@ - r?
Comment 9 WebKit Review Bot 2012-10-16 09:52:30 PDT 
Comment on attachment 168812 [details]
Patch

Clearing flags on attachment: 168812

Committed r131466: <http://trac.webkit.org/changeset/131466>
Comment 10 WebKit Review Bot 2012-10-16 09:52:33 PDT 
All reviewed patches have been landed.  Closing bug.