Correcting the issue number as #3961...
On Thursday 18 August 2011 08:32 PM, Prabhu Gnana Sundar wrote:
Hi all,
With reference to my earlier discussions in
http://svn.haxx.se/dev/archive-2011-07/0432.shtml
I am attaching a python script which would find the bogus mergeinfo
and fix it.
This script would eventually look for the location segments of the
source paths against the respective revision
ranges present in the mergeinfo. With respect to the output of the
location segments it would create a new mergeinfo
and store it in the ".newmergeinfo" file in the path from where the
python script was run. The hash of this file would
be in the ".hashfile" in the same path.
By default, the script would not fix the new changes in the working
copy. Fixing the new mergeinfo to the working copy
can be achieved by passing the "--fix" option to the script.
Also, this script can also authenticate against the self-signed ssl
servers.
I tested this script on our asf subversion repo found a few bogus
mergeinfo such as in the "tree-conflicts" branch:
/subversion/branches/tree-conflicts with 872524-873154,868290-872329
The above should be correct mergeinfo since it was not at all present
in the revision range: r872330-872524
Also, I ran this script against the
https://ctf.open.collab.net/svn/repos/svnedge/trunk/console path and
found a bogus mergeinfo. (it has a self-signed cert)
Please share your thoughts...
Thanks and regards
Prabhu
