[patch] storage refactoring ready to merge?

[Aaron Bentley]

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

Martin Pool wrote:
| Here's an updated version of the "storage" branch, which many people
| have had a hand in by now.  I've reviewed it today and made some small
| changes and will merge it someone else would like to contribute a +1.
|
| The changes (most from other people) include:
|
|  * add LockableFiles grouping
|
|  * store history in a Repository object rather than in the Branch itself
|
|  * remove bzrlib.clone
|
|  * split @needs_read_lock, @needs_write_lock into bzrlib.decorators
|
|  * add IterableFile
|
|  * remove many interfaces from Branch that don't (or no longer)
|    properly belong there

I wish I could get a diff of just the changes that had been made since I
was working on this.  At least if I had your branch, my version of
repository.py would be compared with your version.

A couple of things:

This fixme is wrong.  I think it may always have been wrong.

# FIXME: At the moment locking the Branch locks both the repository and the
# control files, representing the two aspects currently controlled by one
# object.  However, they currently both map to the same lockfile.


~        # IterableFile would not be needed if Transport.put took iterables
~        # instead of files.  ADHB 2005-12-25 # RBC 20060103 surely its
not needed anyway, with codecs transcode
~        # file support ?

I could find things that took unicode and wrote binary, and I could find
things that read binary and produced unicode, but I was unable to find a
way to read unicode and produce binary.

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

iD8DBQFDxxDg0F+nu1YWqI0RAuyXAJ4wc7QM4y4i0Ey7Y2RYysZazs2ITACfVAHO
2xzdYSfX+kNsKedzgH84FIU=
=1nq9
-----END PGP SIGNATURE-----