On Fri, Nov 7, 2008 at 11:07 AM, Julian Foad <julianfoad_at_btopenworld.com> wrote:
> On Fri, 2008-11-07 at 10:07 -0500, Paul Burba wrote:
>> On Fri, Nov 7, 2008 at 9:21 AM, C. Michael Pilato <cmpilato_at_collab.net> wrote:
>> > Paul Burba wrote:
>> >> Reintegrate failed but not in the way expected
>> >> EXPECTED STDOUT (regexp):
>> >> (svn: Reintegrate can only be used if the revisions previously merged
>> >> from the reintegrate target to '.*A_COPY' are the same, but there are
>> >> differences:
>> >> )|( A_COPY
>> >> )|( /A:2-12
>> >> )|( A_COPY/D
>> >> )|( /A/D:2-7,9-12
>> >> )|(
>> >> )
>> >> ACTUAL STDOUT:
>> >> subversion/libsvn_client/merge.c:7226: (apr_err=195016)
>> >> svn: Reintegrate can only be used if the revisions previously merged
>> >> from the reintegrate target to
>> >> 'svn://127.0.0.1:1295/svn-test-work/repositories/merge_tests-126/A_COPY'
>> >> are the same, but there are differences:
>> >> A_COPY
>> >> /A:2-12
>> >> A_COPY/D
>> >> /A/D:2-7,9-12
>> >>
>> >
>> > Could it be because of this bit of the error output?
>> >
>> > subversion/libsvn_client/merge.c:7226: (apr_err=195016)
>>
>> It's a miracle! I can see now, you've cured me! :-\ Fixed in r34094.
>
> r34094:
> [[[
> @@ -14669,7 +14669,8 @@ def reintegrate_with_subtree_mergeinfo(s
> "|( /A/D:2-7,9-12\n)"
> - "|(\n)",
> + "|(\n)"
> + "|(.*apr_err.*)", # In case of debug
> build
> None,
> ]]]
>
> Eww! Do we really have no better way of ignoring that bit of an error
> report? This is the only place in the whole test suite where "apr_err"
> is mentioned. How do all the other tests that check for error messages
> cope with it?
We cope with it everywhere else by not really caring about the other
lines in the error. See this bit from the original diff:
> + # The actions.run_and_verify_* methods are happy if one line of the error
> + # matches the regex, but we want to know that the error actually provides
> + # specific information about the paths that are stopping --reintegrate from
> + # working. So we will pass the stderr to svntest.verify.verify_outputs()
> + # ourselves, but as the 'actual_stdout' argument, that way each line of
> + # error must match the regex.
Usually we care only that an error was raised or about one part of one
line of the error. This case is unusual it that we (well I anyway)
care a bit more than usual about all the lines in the error.
Paul
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe_at_subversion.tigris.org
For additional commands, e-mail: dev-help_at_subversion.tigris.org
Received on 2008-11-07 20:03:48 CET