Am 28.04.2012 14:00, schrieb Stefan Fuhrmann:
Philip Martin wrote:
Philip Martin<philip.mar...@wandisco.com> writes:
On my Linux machine svnserve with threading and revprop caching causes
authz_tests 7 to FAIL:
$ subversion/svnserve/svnserve --cache-revprops yes -Tdr
subversion/tests/cmdline
and
$ ./authz_tests.py --url=svn://localhost 7
Getting lots of failures now:
FAIL: authz_tests.py 7: test authz for log and tracing path changes
FAIL: diff_tests.py 38: diff with svn:author that has< in it
FAIL: log_tests.py 11: svn log --xml must not print empty date elements
FAIL: log_tests.py 25: test revprop retrieval
FAIL: prop_tests.py 30: post-revprop-change hook
FAIL: prop_tests.py 34: test revprop atomicity guarantees of libsvn_ra
FAIL: svnadmin_tests.py 10: 'setlog' and 'setrevprop', bypassing hooks'
FAIL: svnrdump_tests.py 42: svnrdump renumbers dropped revs in
mergeinfo
FAIL: svnrdump_tests.py 43: don't drop mergeinfo revs in incremental
svnrdump
and more, I stopped the tests at that point.
Revprop caching doesn't seem to work at all with threading:
subversion/svnserve/svnserve --cache-revprops yes -Tdr .
svnadmin create repo
ln -s /bin/true repo/hooks/pre-revprop-change
printf "[general]\nanon-access = write"> repo/conf/svnserve.conf
svn ps -r0 --revprop xxx yyy svn://localhost/repo
svn pg -r0 --revprop xxx svn://localhost/repo
The propset reports "property 'xxx' set on repository revision 0" but
the propget return nothing. Using ra_local works:
svn pg -r0 --revprop xxx file://`pwd`/repo
yyy
So the write is happening but the caching is failing. Using a
non-threaded svnserve appears to work.
Hey Philip,
Thanks for testing and providing the reproduction recipes!
With r1331763, all these issues should be fixed. Only the
hotcopy tests still need to be fixed / looked into.
Forget that comment about hotcopy - that is a issue
Daniel is looking into. With my latest svnadmin commit
all tests should now pass.
-- Stefan^2.
