| Summary: | IndexedDB: call abort handler when there are problems committing | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Product: | WebKit | Reporter: | Alec Flett <alecflett> | ||||||
| Component: | New Bugs | Assignee: | Alec Flett <alecflett> | ||||||
| Status: | RESOLVED FIXED | ||||||||
| Severity: | Normal | CC: | dgrogan, ojan, webkit.review.bot | ||||||
| Priority: | P2 | ||||||||
| Version: | 528+ (Nightly build) | ||||||||
| Hardware: | Unspecified | ||||||||
| OS: | Unspecified | ||||||||
| Attachments: |
|
||||||||
|
Description
Alec Flett
2012-05-07 16:07:28 PDT
Created attachment 140759 [details]
Patch
LGTM Comment on attachment 140759 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=140759&action=review > Source/WebCore/ChangeLog:8 > + No new tests. (OOPS!) You'll have to remove the OOPS and re-upload. The reasoning from my similar patch applies here (https://bugs.webkit.org/attachment.cgi?id=140156&action=review) No new tests. We think the only way this will happen is if the leveldb store gets corrupted, either due to IO problems or coding error. We're working on a pyauto test that corrups the db in the background but can't do that from DRT. Created attachment 141009 [details]
Patch
ojan - r? No tests here, as explained in the changelog - this is going to help us reproduce a problem we're seeing in the wild, but this is quite safe otherwise. (and all our existing tests already basically call commit()) Comment on attachment 141009 [details]
Patch
oops, forgot the cq? - jsbell@
er rather, dgrogan@ - cq? Comment on attachment 141009 [details] Patch Clearing flags on attachment: 141009 Committed r116562: <http://trac.webkit.org/changeset/116562> All reviewed patches have been landed. Closing bug. |