>> Should I file a separate bug to fix tests that depend on openCacheEntry seeing >> that you've resolved this bug as FIXED? > oh, tests, yeah, separate bug please References: Bug 737615 - Remove use of synchronous cache API from unit tests.

> References: > Bug 737615 - Remove use of synchronous cache API from unit tests. http://hg.mozilla.org/mozilla-central/rev/72019f633129

http://mxr.mozilla.org/comm-central/search?string=.openCacheEntry(&find=/suite/ Finds two files; http://mxr.mozilla.org/comm-central/source/suite/common/places/tests/unit/test_clearHistory_shutdown.js http://mxr.mozilla.org/comm-central/source/suite/modules/test/browser_sanitizer.js

Comment on attachment 682869 [details] [diff] [review] Patch for test_clearHistory_shutdown.js This patch ports this change http://hg.mozilla.org/mozilla-central/diff/72019f633129/browser/components/places/tests/unit/test_clearHistory_shutdown.js to the SeaMonkey version of this test. No changes to the FF fix were necessary, I've checked that the test passes. I'll remove the bogus white space change before + var storeCacheListener = { before pushing that patch.

Comment on attachment 682869 [details] [diff] [review] Patch for test_clearHistory_shutdown.js Pushed: https://hg.mozilla.org/comm-central/rev/a6f6ed53b376

I need to investigate what happens on the testboxen, this test crashes now. Stack from Linux testbox: 0 libmozalloc.so!mozalloc_abort(char const*) [mozalloc_abort.cpp : 23 + 0x0] 1 libxul.so!NS_DebugBreak_P [nsDebugImpl.cpp : 423 + 0x5] 2 libxul.so!mozilla::storage::StatementCache<mozIStorageStatement>::CreateStatement(nsACString_internal const&) [StatementCache.h : 125 + 0x21] 3 libxul.so!mozilla::storage::StatementCache<mozIStorageStatement>::GetCachedStatement(nsACString_internal const&) [StatementCache.h : 59 + 0x10] 4 libxul.so!mozilla::places::Database::GetStatement(nsACString_internal const&) const [Database.h : 172 + 0xe] 5 libxul.so!already_AddRefed<mozIStorageStatement> mozilla::places::Database::GetStatement<245>(char const (&) [245]) const [Database.h : 151 + 0xc] 6 libxul.so!nsNavBookmarks::FetchItemInfo(long long, mozilla::places::BookmarkData&) [nsNavBookmarks.cpp : 1433 + 0x22] 7 libxul.so!nsNavBookmarks::RemoveFolderChildren(long long) [nsNavBookmarks.cpp : 1142 + 0x12] 8 libxul.so + 0x1776d0b 9 libxul.so!CallMethodHelper::Invoke() [XPCWrappedNative.cpp : 3139 + 0xf] [...]

Line 125 is NS_ERROR, hopefully there should be a message earlier in the log?

This is the console output from the testbox before the crash.

OK, so we have ###!!! ASSERTION: The statement 'SELECT b.id, h.url, b.title, b.position, b.fk, b.parent, b.type, b.dateAdded, b.lastModified, b.guid, t.guid, t.parent FROM moz_bookmarks b LEFT JOIN moz_bookmarks t ON t.id = b.parent LEFT JOIN moz_places h ON h.id = b.fk WHERE b.id = :item_id' failed to compile with the error message ''.: 'Error', file ../../../dist/include/mozilla/storage/StatementCache.h, line 125 So, I guess we're trying to compile this statement when we shouldn't? (A pity there's no JS stack...)

Ping mcsmurf. I don't suppose you've made any progress on this?

This test magically(?) seems to pass now and not crash (from beta builds): TEST-INFO | /builds/slave/test/build/xpcshell/tests/suite/common/places/tests/unit/test_clearHistory_shutdown.js | running test ... TEST-PASS | /builds/slave/test/build/xpcshell/tests/suite/common/places/tests/unit/test_clearHistory_shutdown.js | test passed (time: 797.629ms) TEST-INFO | e:\builds\slave\test\build\xpcshell\tests\suite\common\places\tests\unit\test_clearHistory_shutdown.js | running test ... TEST-PASS | e:\builds\slave\test\build\xpcshell\tests\suite\common\places\tests\unit\test_clearHistory_shutdown.js | test passed (time: 891.000ms) Resolving bug as fixed then. Not really sure in which release this was fixed, so setting no target milestone.

> so setting no target milestone. I'll just pick one at random then ;)