Jani Averbach wrote:
> On 2004-11-03 04:35+0100, brane@xbc.nu wrote:
>>FAIL: merge_tests.py 16: merge with funny characters (issue #1995)
>
> This patch fixes that failure. Do we like to apply it only to the
> 1.1.x branch or also for trunk? The dry run works just fine on trunk.
I don't see why we would want to disable a working regression test on the trunk.
If I understand you correctly, there is some aspect of merging that works on
trunk, but is broken - and fails a regression test - in 1.1.x. If that is the
case, then how did the branch get into that state?
a) A new bug was ported onto the branch;
b) A new regression test was ported onto the branch;
c) The test has always been failing on 1.1.x but has only now been noticed.
None of these should have happened - as any code being ported onto the branch
should have been tested first - but obviously things can go wrong. Unless
there are better reasons or I am misunderstanding, I feel that whatever was
done to the branch to cause this failure should be undone until a fully working
patch is available to be ported from the trunk.
If, for whatever reasons, the branch is going to have to live with this bug and
we cannot revert the patch that caused it to start failing, then it would be
reasonable to disable this (part of a) regression test on the branch only, but
there is still no reason to disable it on the trunk (except for the fact that
the general procedure is to make a change on trunk and then port it to the
branch - but that procedure is based on the premise that a change is equally
applicable to trunk and to branch, which is not the case here).
- Julian
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Wed Nov 3 19:21:54 2004