|Submitter||Diff||Changes||Open Issues||Last Updated|
|Error loading review requests:|
58 bytes, text/x-review-board-request
|Details | Review|
Items in the bookmark's guidMap() are created via |new String()| so that additional properties can be added to the object. Now that PlacesSyncUtils exists, we validate GUIDs via PlacesUtils.isValidGuid() which rejects such objects. Best I can tell, this breaks our duplicate handling. Sadly there are no tests for this in the tree currently, but I'm working on some in bug 1276823.
Comment on attachment 8781346 [details] Bug 1295410 - ensure we don't pass string objects for GUIDs to PlacesSyncUtils. https://reviewboard.mozilla.org/r/71762/#review69352 Ah, good catch! I fixed another occurence of this in http://searchfox.org/mozilla-central/rev/9ec085584d7491ddbaf6574d3732c08511709172/services/sync/modules/engines/bookmarks.js#789, but not this one.
Attachment #8781346 - Flags: review?(kcambridge) → review+
Pushed by firstname.lastname@example.org: https://hg.mozilla.org/integration/autoland/rev/714f26001d3e ensure we don't pass string objects for GUIDs to PlacesSyncUtils. r=kitcambridge
Also, this seems familiar; I think it's come up before for Thom, too. It'd be great to use POJOs instead of relying on this JS quirk. Filed bug 1295510.
Status: NEW → RESOLVED
Last Resolved: 2 years ago
status-firefox51: --- → fixed
Resolution: --- → FIXED
Target Milestone: --- → Firefox 51
You need to log in before you can comment on or make changes to this bug.