On Tue, Aug 30, 2011 at 11:36 AM, Philip Martin
<philip.mar...@wandisco.com> wrote:
> Johan Corveleyn <jcor...@gmail.com> writes:
>
>> Status against revision: 2
>> =============================================================
>> Expected 'gamma' and actual 'gamma' in status tree are different!
>> =============================================================
>> EXPECTED NODE TO BE:
>> =============================================================
>> * Node name: gamma
>> Path: svn-test-work\working_copies\depth_tests-18-1\A\D\gamma
>> Contents: None
>> Properties: {}
>> Attributes: {'status': 'M ', 'wc_rev': '1'}
>> Children: None (node is probably a file)
>> =============================================================
>> ACTUAL NODE FOUND:
>> =============================================================
>> * Node name: gamma
>> Path: svn-test-work\working_copies\depth_tests-18-1\A\D\gamma
>> Contents: None
>> Properties: {}
>> Attributes: {'status': ' ', 'wc_rev': '1'}
>> Children: None (node is probably a file)
>
> Do you still have the working copy? The file hasn't been committed
> (still at r1) but is unmodified. Perhaps it's just that the timestamps
> match and the text modification is still present?
Hi Philip,
Thanks for your feedback.
No, I'm afraid I don't have the working copy anymore, so there's no
way to be sure.
I should have been more careful, but when I saw that test failing, I
just couldn't believe it, because I had successfully run the entire
test suite on the 1.7.x branch just before that. So I decided to rerun
the depth_tests, just to make sure it failed. Which wasn't very smart,
because then the test succeeded, and the broken working copy was gone
:-(.
Anyway, it may be a reasonable hypothesis. I'm not sure if it will
ever happen again on my system. If this is really the cause, does that
mean that it's a problem for Subversion, or that it's more of a
test-suite problem (that it should sleep some time to make sure the
timestamps differ, or something like that)?
--
Johan
