Re: svn log -g problems

C. Michael Pilato wrote:
> Mark Phippard wrote:
>> On Mon, Mar 10, 2008 at 11:53 AM, Mark Phippard <markphip_at_gmail.com>
>> wrote:
>>
>>> I saw you committed something, but the behavior seems the same for me.
>>> At least in the sense that when running the API over a range only the
>>> first item that was a merge returns mergeinfo output. For example, I
>>> have the 1.5 branch built @ r29823 (current HEAD). If I run this
>>> command on that working copy:
>>
>> Just as a followup, I have a repository I access via ra_local that
>> does seem to show mergeinfo beyond the first merge. So maybe this is
>> a problem via certain ra methods.
>>
>> I also notice that the merged revisions seem to count towards the
>> --limit, and I do not think they should. I imagine that is debatable,
>> but I only expect --limit to apply to the "main revisions".
>
> Here's what I get running ra-neon, ra-local, and ra-serf against my
> local sync of our repository: http://paste.lisp.org/display/57133

I've failed to determine -- even under a debugger -- why 'svn log -g' is
trying to lookup history for a non-existent path. But with a local mirror
of our source code repository, it's pretty easy to trigger the error:

   svn log -vq -g --limit 20 \
       file:///usr/local/svn/subversion/branches/1.5.x_at_29768

-- 
C. Michael Pilato <cmpilato_at_collab.net>
CollabNet   <>   www.collab.net   <>   Distributed Development On Demand