[MERGE REVIEW] Revert destroys file contents produced by merge

[Aaron Bentley]

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

Robert Collins wrote:
> On Tue, 2006-02-28 at 23:22 -0500, Aaron Bentley wrote:
> I'm not sure that test_merge_revert belongs in test_inv.py...

That's where the existing tests of revert were.

> Probably its a working tree format specific test - all working tree
> implementations should behave the same way ? (or should only format-3
> working trees (those in metadirs) behave like this.

All implementations should behave the same way.

> If its all working trees it should be tests/workingtree_implementations;
> if its just format 3 trees, it should be in tests/test_workingtree.py
> and there should be a matching test for format 2 trees that there
> behaviour is unaltered from the 0.7 release.

I would like a better explanation of these tests that are run across
varying formats/transports, at some point.  For now, I have moved the
test into tests/workingtree_implementations/test_workingtree.py.
Updated patch attached.

> Whats the impact of a 0.7 client working with a tree that has had this
> merge-hashes file written into it? 

It will ignore it.  Merge will not update the file, and revert will pay
no attention to it.  This will not cause any problems if a 0.8 client is
used on that tree later, except that if a 0.7 client does a merge, the
0.8 client will preserve those changes (in backups, or by unversioning)
when it reverts.

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

iD8DBQFEDFW50F+nu1YWqI0RAmQHAJsFu8oFj0R52tm6E+SCZgzPf1aq3ACffEFr
MW0Q+wdVCZ7iz8aRdf2SdJQ=
=s/0C
-----END PGP SIGNATURE-----
-------------- next part --------------
A non-text attachment was scrubbed...
Name: merge-modified2.patch
Type: text/x-patch
Size: 10318 bytes
Desc: not available
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20060306/dddd0e3d/attachment.bin