Saw these failing on Windows, but didn't want to make to hold up landing bug 1131362, since it's blocking pretty much all work. Two tests are failing on my Windows box with: 0:01.82 LOG: Thread-22 ERROR NS_ERROR_FILE_IS_LOCKED: Component returned failure code: 0x8052000e (NS_ERROR_FILE_IS_LOCKED) [nsIFile.remove] removeDB@d:/mozilla/central/obj-i686-pc-mingw32/_tests/xpcshell/browser/components/readinglist/test/xpcshell/test_SQLiteStore.js:22:6 _execute_test@d:\mozilla\central\testing\xpcshell\head.js:557:15 This is a classic issue on Windows, so I wasn't overly concerned about it.

P2 because this is a new test for our feature.

Surely it's better to disable them on Windows only... In the meantime I'll try to figure out what the problem is since this is news to me.

This should be the proper fix. Try is green so far: https://treeherder.mozilla.org/#/jobs?repo=try&revision=bff1dae02a52

Comment on attachment 8575568 [details] [diff] [review] wait for the DB to close before trying to remove the file Review of attachment 8575568 [details] [diff] [review]: ----------------------------------------------------------------- ::: browser/components/readinglist/test/xpcshell/test_ReadingList.js @@ +22,5 @@ > Assert.ok(gList); > gDBFile.append(gList._store.pathRelativeToProfileDir); > + do_register_cleanup(function* () { > + // Wait for the store to close its connection to the database. > + yield gList._store.destroy(); Isn't it safer for this to be calling gList.destroy()?

Should this bug be prioritized as a P1 for the campaign release?

(In reply to Marco Mucci [:MarcoM] from comment #5) > Should this bug be prioritized as a P1 for the campaign release? We discussed this during the triage meeting - we *really* want it, but can theoretically ship without. So P2.

Comment addressed. Of course the tree is closed.