[1.13][merge] set default format to 1.9 and tweak remote branch stacking

[John Arbash Meinel]

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

Martin Pool wrote:
> This
> 
>  * sets the default repository format to 1.9, which is straightforward
> but needs a few changes to blackbox tests that include the format
> number
> 
>  * adds reprs for some objects
> 
>  * changes RemoteBranchFormat so that it declares it may support
> stacking, unless the specified underlying format is known not to.  I'm
> not sure that's totally right in the case of eg a remotebranch to
> which we have not yet connected and so know nothing about the format,
> but it's closer.
> 
>  * adds an assertLength test assertion to help with eg checks about
> how many rpcs are made
> 
> Unfortunately this still has some test failures to do with stacking on
> remotebranches and choosing the right format, but I'm going to share
> it now.


I would really like to see this land for 1.13, its sort of a shame we
didn't think about it until now.

This change to the docs is no longer correct:
- -* If you are using bzr-svn to interoperate with a Subversion
- -  repository, use 1.9-rich-root.
- -
- -* If you are working on a project with big trees (5000+ paths)
- -  or deep history (5000+ revisions), use 1.9.
- -
- -* Otherwise, use the default format - it is good enough for
- -  most projects.
+* Otherwise, use 1.9-rich-root, which is the default and has
+  good performance on large trees and histories.


I know you were thinking to make rich-root the default, and then changed
your mind.

=== modified file 'bzrlib/remote.py'
- --- bzrlib/remote.py	2009-03-06 02:21:00 +0000
+++ bzrlib/remote.py	2009-03-06 09:54:31 +0000
@@ -1567,13 +1567,18 @@

...
+    def supports_stacking(self):
+        if self._custom_format:
+            return self._custom_format.supports_stacking()
+        else:
+            # Not every remote format supports stacking, but they
might; this
+            # is maybe not quite right.
+            return True
+

Shouldn't this be doing some sort of '_ensure_real()' dance?


If it was clearer to me what needs to happen wrt stacking, I would
probably clean this up and submit it myself. I'll see how today goes, as
it would be nice to get it into 1.13rc1.

BB:tweak

John
=:->
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (Cygwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEYEARECAAYFAkmxMOEACgkQJdeBCYSNAAOPhwCfVXiRsml3umSnJXhRmCQBITED
YNEAoLJvixzrS//Xx2GAuAJKNsV2TgKS
=qXcl
-----END PGP SIGNATURE-----