"bzr revert" leaves a mess

[Aaron Bentley]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Robey Pointer wrote:
> Last night, I ran into a merge conflict and I was trying to get more 
> debugging info out of it for Aaron, so I did:
> 
> $ bzr uncommit
> $ bzr revert
> 
> and tried the merge again.  This time, of course, I got dozens of 
> conflicts, though I didn't know why at the time.  Turns out that "bzr 
> revert" leaves behind any files created by the merge, as well as 
> creating a bunch of new *~ and *.moved files (and folders!).
> 
> I can understand the motivation ("bzr revert" used to erase newly- added
> files, which was much worse) but I think we should do one of:
> 
> 1. When reverting a "merge" operation, erase added files.
> 
> 2. Make an option "bzr revert --clean" which removes added files.

How about this:
When a merge is applied, we note the hashes of the new/modified files
somewhere in .bzr.

When we revert,
 - If a file is new, but it matches its merge hash, we delete it
 - If a file is modified, but it matches its merge hash, we do not make
a backup.

Aaron

> 
> robey
> 
> 

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFD/LEy0F+nu1YWqI0RAgvzAJ4wcRGZVZ2DxxErxWT2wj8wAWKmAACfZ+uD
WHG4uExnENzBVqaU6BQs3ys=
=dd8h
-----END PGP SIGNATURE-----