a6e940645d5e68009caa0ef1645ef3821299f325: lose some of the more excessive sweep-related debug
Andrew Sutherland <asutherland@asutherland.org> - Sun, 21 Sep 2008 10:28:13 -0700 - rev 937
lose some of the more excessive sweep-related debug
293a9b25dc5190de905c839bd46a6a11208bc344: get rid of some debug
Andrew Sutherland <asutherland@asutherland.org> - Sun, 21 Sep 2008 08:15:10 -0700 - rev 936
get rid of some debug
26164ab1d4041e5867957f1e3fbf58b44e32b71e: clean up after ourselves with the emitter.
Andrew Sutherland <asutherland@asutherland.org> - Sun, 21 Sep 2008 06:56:47 -0700 - rev 935
clean up after ourselves with the emitter.
13592e09f549e323eab7719286d45b36d2f99148: things seem to work, but the unit tests are so very angry.
Andrew Sutherland <asutherland@asutherland.org> - Sun, 21 Sep 2008 01:16:22 -0700 - rev 934
things seem to work, but the unit tests are so very angry.
eb07537051591a8ed2254f56804f20f64e2f7d8a: Add concept of 'unique value' as an alternate lookup mechanism for
Andrew Sutherland <asutherland@asutherland.org> - Sat, 20 Sep 2008 23:36:26 -0700 - rev 933
Add concept of 'unique value' as an alternate lookup mechanism for collections. This allows us to avoid creation of duplicate identities due to the lookup currently being synchronous and the insert being async. It does assume the cache/collections are sufficient though. We probably need some concept of locking the objects into memory until their transaction is committed.
9082e6d74f2d06a0a0556585ec78174fbe276cc0: indexer status commit; want a separate commit for some collection changes.
Andrew Sutherland <asutherland@asutherland.org> - Sat, 20 Sep 2008 22:41:26 -0700 - rev 932
indexer status commit; want a separate commit for some collection changes.
c2df7b8472375d2cff1684a8b6aec7f15355840f: status commit, private propagation so as to not break peoples.
Andrew Sutherland <asutherland@asutherland.org> - Sat, 20 Sep 2008 05:00:13 -0700 - rev 931
status commit, private propagation so as to not break peoples.
1f547582f25ee366fc65d754bb93b31a69c4d5e0: status commit for computer change.
Andrew Sutherland <asutherland@asutherland.org> - Fri, 19 Sep 2008 19:39:31 -0700 - rev 930
status commit for computer change.
63628cda0155ba3ab503a8658f8e280d3ab84db1: no need for _extractOriginalSubject abstraction
Andrew Sutherland <asutherland@asutherland.org> - Fri, 19 Sep 2008 15:06:09 -0700 - rev 929
no need for _extractOriginalSubject abstraction
74b3e2a7887f956c3d6c6b16e894f12d28cd16a5: indexing changes status commit; not working! (I'm not going to push until the
Andrew Sutherland <asutherland@asutherland.org> - Fri, 19 Sep 2008 04:03:41 -0700 - rev 928
indexing changes status commit; not working! (I'm not going to push until the working commit gets in...)
55a9fce200d63c33e59d3becc1292d7649b6590a: Support mbox injection in addition to fakeserver injection for testing.
Andrew Sutherland <asutherland@asutherland.org> - Wed, 17 Sep 2008 22:09:47 -0700 - rev 927
Support mbox injection in addition to fakeserver injection for testing.
9816bf270ca211822ac213ae4cf194c2af460bf2: bad meta-slash; wrong completion.
Andrew Sutherland <asutherland@asutherland.org> - Sun, 14 Sep 2008 22:43:07 -0700 - rev 926
bad meta-slash; wrong completion.
12dd991ee463b3df7ff229a6a4507f49baf75c2f: whoops, gloda collection manager is still lacking on the id-deletion front.
Andrew Sutherland <asutherland@asutherland.org> - Sun, 14 Sep 2008 20:10:13 -0700 - rev 925
whoops, gloda collection manager is still lacking on the id-deletion front.
739c3b36c1554e49ce7659fd7e76bae4715781c3: bad redundant let, bad!
Andrew Sutherland <asutherland@asutherland.org> - Sun, 14 Sep 2008 20:01:35 -0700 - rev 924
bad redundant let, bad!
308e8ccddd6b7b949b6af42cc8087eb27a4d3f74: * proper schema change, database will be purged
Andrew Sutherland <asutherland@asutherland.org> - Sun, 14 Sep 2008 00:51:09 -0700 - rev 923
* proper schema change, database will be purged * added per-message full-text subject indexing (not just initial conversation subject) * added per-message actual file attachment full-text indexing * added limited (but workable) heuristic to detect "real" attachments (their name isn't just Part 1.N...)
97ad6682ea4cb2cd825d7b74eec08b211098fe79: * improve shutdown logic
Andrew Sutherland <asutherland@asutherland.org> - Sat, 13 Sep 2008 23:36:28 -0700 - rev 922
* improve shutdown logic * happen at quit-application, not xpcom-shutdown * actually finalize everyone who should be finalized * just in case, be sure to close out any pending transaction * actually call datastore shutdown from indexer * fix blatantly foolish partly-done refactoring in attribute def lookup. * that problem necessitated a db version bump to purge duplicate attributes * we also had some schema changes that it's good to get in there too...
f36de63d524afc85d6b98bd119b87942026fd48a: deprecate the explain method; we've already gutted the explanations themselves.
Andrew Sutherland <asutherland@asutherland.org> - Sat, 13 Sep 2008 23:05:09 -0700 - rev 921
deprecate the explain method; we've already gutted the explanations themselves.
4f91d4688214161e5da17625fc01d5f74e0be2e4: introduce a helper function that hopefully will deal with the rare, but
Andrew Sutherland <asutherland@asutherland.org> - Sat, 13 Sep 2008 20:53:02 -0700 - rev 920
introduce a helper function that hopefully will deal with the rare, but expected SQLITE_BUSY failures from our sync/async setup. We want the request to keep trying until the async transaction in the process of completion completes and our sync query can go through.
d101bf1e78316d204b7cb6ffbcf327da44301e1b: use transactions again
Andrew Sutherland <asutherland@asutherland.org> - Fri, 12 Sep 2008 19:08:31 -0700 - rev 919
use transactions again
4cb51eaaea150fb1d9359f6524121901b0f6e4e6: fix foolish typo in async query result handling
Andrew Sutherland <asutherland@asutherland.org> - Fri, 12 Sep 2008 18:15:57 -0700 - rev 918
fix foolish typo in async query result handling
(0) -300 -100 -50 -20 +20 +50 +100 +300 +1000 +3000 +10000 +30000 tip