Rev 5764: Merge config-abstract-store into config-concrete-stores in file:///home/vila/src/bzr/experimental/config/

[Vincent Ladeuil]

At file:///home/vila/src/bzr/experimental/config/

------------------------------------------------------------
revno: 5764 [merge]
revision-id: v.ladeuil+lp at free.fr-20110407092846-miux7haa13hb1q95
parent: v.ladeuil+lp at free.fr-20110407074453-9guexqsmrisvvqcn
parent: v.ladeuil+lp at free.fr-20110407092835-1eorign73b8h6p0i
committer: Vincent Ladeuil <v.ladeuil+lp at free.fr>
branch nick: config-concrete-stores
timestamp: Thu 2011-04-07 11:28:46 +0200
message:
  Merge config-abstract-store into config-concrete-stores
modified:
  bzrlib/tests/test_config.py    testconfig.py-20051011041908-742d0c15d8d8c8eb
-------------- next part --------------
=== modified file 'bzrlib/tests/test_config.py'
--- a/bzrlib/tests/test_config.py	2011-04-07 07:44:53 +0000
+++ b/bzrlib/tests/test_config.py	2011-04-07 09:28:46 +0000
@@ -2076,8 +2076,14 @@
         store.save()
 
     # FIXME: We should adapt the tests in TestLockableConfig about concurrent
-    # writes, for now, I'll just rely on using the same code (copied, but
-    # pretty trivial) -- vila 20110-04-06
+    # writes. Since this requires a clearer rewrite, I'll just rely on using
+    # the same code in LockableConfigObjStore (copied from LockableConfig, but
+    # trivial enough, the main difference is that we add @needs_write_lock on
+    # save() instead of set_user_option() and remove_user_option()). The intent
+    # is to ensure that we always get a valid content for the store even when
+    # concurrent accesses occur, read/write, write/write. It may be worth
+    # looking into removing the lock dir when it;s not needed anymore and look
+    # at possible fallouts for concurrent lockers -- vila 20110-04-06
 
 
 class TestConfigObjStore(tests.TestCaseWithTransport):