[MERGE] Get rid of revision_history() call during copy_content_into.

[John Arbash Meinel]

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

Andrew Bennetts wrote:
> Aaron Bentley wrote:
>> -----BEGIN PGP SIGNED MESSAGE-----
>> Hash: SHA1
>>
>> Andrew Bennetts wrote:
>>> This removes a call to revision_history() that was occurring during a
>>> copy_content_into.  In particular this was affecting RemoteBranch.  This
>>> will penalise BzrBranch5 a little, but it's old enough that we probably
>>> don't mind.
>> bb:resubmit
>>
>> This doesn't make sense to me.  If it's affecting any branches, it's
>> affecting RemoteBranch only, because it's optimized for each branch type.
> 
> It can affect foreign branches that subclass Branch as well.  I'd rather that
> slow, deprecated code was the not the default case.
> 
>> Please optimize RemoteBranch instead of pessimizing Branch5.
> 
> Ok.  Our inheritance heirarchy is pretty whack here:
> 
>   Branch
>    |   \______________________
>    |                          \
>   BzrBranch (format 4)       RemoteBranch
>    |
>   BzrBranch5
>    |
>   BzrBranch7
>    |
>   BzrBranch6
> 
> 4, 5, 7, 6.
> 
> So fixing Branch to default to the modern interfaces, while allowing format 4/5
> to use revision-history, and still allowing 6/7 to use the modern facilities, is
> a bit ugly.  This patch does that.
> 
> We probably want to dramatically restructure this code at some point, but that's
> out-of-scope for this patch, unsurprisingly.
> 
> This version of the patch also adds a test that shows that RemoteBranch no
> longer does a Branch.revision_history RPC.
> 
> -Andrew.
>   
> 

BB:tweak

I'll let Aaron bring up any concerns he has, but this looks fine to me.

John
=:->

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

iEYEARECAAYFAkkdGoIACgkQJdeBCYSNAAMIUQCfQEc/MiDc9SuUfTmUcStMmkre
2n0An1O0ZAodTssXvOeWX/nhjwT14A9+
=31qM
-----END PGP SIGNATURE-----