659112 - Surprise test failure in test_bookmark_engine.js, test_corrupt_keys.js

Status: RESOLVED FIXED

(Firefox :: Sync, defect)

Description

[Richard Newman [:rnewman]]

"constraint failed" when changing GUID, line 55. Looks like our 'random' GUID is no longer acceptable!

Comment 1

[Richard Newman [:rnewman]]

Attachment: Proposed patch. v1

Comment 2

[Richard Newman [:rnewman]]

Part two is a lack of HMAC failures in test_corrupt_keys, presumably because our lastSync isn't starting afresh at 0...

Comment 3

[Richard Newman [:rnewman]]

Attachment: Part 2. v1

Fix this in two places: firstly by refreshing a single server item, and secondly by just setting lastSync to 0. That matches up with the intent of the test.

Comment 4

[Philipp von Weitershausen [:philikon]]

Attachment: fix root cause in test harness

Comment 5

[Philipp von Weitershausen [:philikon]]

Comment on attachment 534579 [details] [diff] [review]
fix root cause in test harness

I should explain this: basically, since we call do_get_profile(), we already have a ProfD directory, we don't need to provide one. If we do and we map it to cwd, it will get shared between test runs and non-clobber builds will have test failures.

Comment 6

[Richard Newman [:rnewman]]

Comment on attachment 534579 [details] [diff] [review]
fix root cause in test harness

Two consecutive all-pass test runs. Looks good.

Comment 7

[Philipp von Weitershausen [:philikon]]

http://hg.mozilla.org/services/services-central/rev/36e0a6facb40

Comment 8

[Philipp von Weitershausen [:philikon]]

http://hg.mozilla.org/mozilla-central/rev/36e0a6facb40