> Jens, could you also try updating mergeinfo.py to set self.fs = None
> and self.repos = None and see if that helps? I've attached an
> (untested) patch which does that.
Hi David,
yes, your patch works perfectly for merge-info and I could apply the
same fix to repository (see attachment).
But unfortunately this one doesn't work for the trac based tests. Within
every trac base suite the first one is fine, but it fails when TearDown
(called from setUp) wants to remove the repository. The stacktrace is
everywhere the same:
subversion\bindings\swig\python\tests\trac\versioncontrol\tests\svn_fs.py",
line 58, in setUp
self.tearDown()
subversion\bindings\swig\python\tests\trac\versioncontrol\tests\svn_fs.py",
line 67, in tearDown
repos.delete(REPOS_PATH)
Jens
------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=462&dsMessageId=1967320
Index: mergeinfo.py
===================================================================
--- mergeinfo.py (revision 37168)
+++ mergeinfo.py (working copy)
@@ -51,6 +51,8 @@
self.rev = fs.youngest_rev(self.fs)
def tearDown(self):
+ self.fs = None
+ self.repos = None
if os.path.exists(REPOS_PATH):
repos.delete(REPOS_PATH)
Index: repository.py
===================================================================
--- repository.py (revision 37168)
+++ repository.py (working copy)
@@ -39,6 +39,10 @@
self.fs = repos.fs(self.repos)
self.rev = fs.youngest_rev(self.fs)
+ def tearDown(self):
+ self.fs = None
+ self.repos = None
+
def test_create(self):
"""Make sure that repos.create doesn't segfault when we set fs-type
using a config hash"""
Received on 2009-04-28 21:07:58 CEST