[Critical] dirstate format use OS file locking

[John Arbash Meinel]

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

John Arbash Meinel wrote:
> Martin Pool wrote:

...

> I don't know of any specific cases were we genuinely *need* a source
> working tree in readonly and a destination workingtree in write, *and*
> they could be the same one.
> 
> The only one off-hand is "bzr merge --uncommitted" which would use 2
> working trees. But I think "bzr merge --uncommitted ." is a user error.
> 
> I'll be looking into it today.


I did find a case where we have this happening, and I'm not positive
what the best fix is. Specifically the merge code likes to open up new
trees for 'this_tree', 'base_tree', and 'other_tree'. And they may be
the same tree. (I specifically see examples where base_tree and
this_tree are the same as part of the bundle tests).

I think the more "correct" thing, is to have the merge code not opening
up its own trees and branches, but instead have these things passed in
from the higher level.

So at this point I'm not sure how to fix it. I'll try and do it the
"correct" way, and see how difficult that is.

John
=:->
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.3 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFF8HNQJdeBCYSNAAMRApciAJ9J6e0Uab9SbTbXmOC5LwMgNQTEpQCgiBXL
VAICSTV06mwsgjUPG7fqEME=
=vnfj
-----END PGP SIGNATURE-----