Bert Huijben wrote:
>> Author: julianfoad
>> Date: Thu Jan 22 18:14:53 2015
>> New Revision: 1653988
>>
>> URL: http://svn.apache.org/r1653988
>> Log:
>> Fix issue #4551 "svnrdump load commits wrong properties, or fails, on a
>> non-deltas dumpfile", by teaching the svnrdump loader to find
Evgeny Kotkov writes:
>> Evgeny Kotkov writes:
>>
>>> The reason why this error is propagated up the stack is that we only examine
>>> the 'ignore_enoent' argument after the first apr_file_remove() call. This
>>> is
>>> racy — if we get a EACCES during the first attempt to remove a file, and t
Philip Martin writes:
>> Here is the patch for this part of the issue. Log message:
>
> I have already committed something, but feel free to commit an
> additional change if you think it better.
I committed a couple of svn_io_remove_file2() tweaks in r1654213 and r1654215.
Regards,
Evgeny Kot
I (Julian Foad) wrote:
> Bert Huijben wrote:
>> I also see new test failures on the two new tests for ra_serf.
>
> It's another real bug in svnrdump. It calls
> commit_editor->change_dir_prop() twice for the same property name, which is
> documented as not allowed. The other RA layers don't mind
I (Julian Foad) wrote:
> I also noticed that configuring with '--enable-ev2-shims' creates the
> same driving order violation, so that several tests would fail against
> RA-serf,
> but since r1654194 they no longer fail.
Correction: tests wouldn't fail on their own in that configuration. What I
5 matches
Mail list logo
