Community technical support mailing list was retired 2010 and replaced with a professional technical support team. For assistance please contact: Pre-sales Technical support via email to sales@march-hare.com.
Hi all,
Since upgrading our CVSNT server to 2.5.0.2382 (we used 2.0.58d until
now) we experience several merge issue.
One of them is merging from a branch to the trunk. It happened twice
that the content of the branch was not entirely merged in the trunk.
Another one that happened today is when a user updated its modified
local copy (merging then changes from another user) :
USER_A start modifying 1.50.2.16
USER_B made some changes and commited them up to 1.50.2.18
So the tree is :
...
|
1.50.2.16
|
1.50.2.17
|
1.50.2.18
with USER_A working on 1.50.2.16
Changes from USER_A and USER_B do not overlap so we don't expect to have
any conflict.
When USER_A updated, CVS told him he was merging changes from USER_B but
in fact, it was not, every changes USER_B made were discarded
(unfortunatly we discovered it after the merged file was commited)
Of course, trying to reproduce the error by rebuilding the file
"hierarchy" in another module failed. Everything work then as expected.
So my questions are :
Am I the only one facing these kind of issues ?
Can it be my setup ? (server is 2.5.0.2382, clients are 2.5.0.2382
through WinCVS)
Can I give you anything to help you solving this ?
Who is making the merge ? the server or the client ? Is it internal
procedure or does it rely on third party merger ?
Is it safe to go back to server 2.0.58d from 2.5.0.2382?
These incident really hurts now because some new developers are pushing
for going to perforce. Of course these 3 issues do not speak in favor of
CVSNT though we are using it for 4 years now without any incident
until recently (which I associate with the upgrade to 2.5.0.2382, but
this might be pure coincidence).
thanks for your great work and the help,
Sebastien