[MERGE] special case the no-args version of revision-info

[John Arbash Meinel]

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

Michael Hudson wrote:
> Michael Hudson wrote:
>> Ian Clatworthy wrote:
>>> Michael Hudson wrote:
>>>> Reading this post:
>>>>
>>>> http://blog.orebokech.com/2008/03/emacs-in-bzr-initial-impressions.html
>>>>
>>>> I was prompted to look at the implementation of version-info.  It seems
>>>> to be a bit whole-history happy: RevisionSpec.in_branch calls
>>>> revsion_history() and the command implementation calls
>>>> get_revid_to_revno_map().
>>>>
>>>> This bundle just makes it really fast for the no-arg case.  Perhaps it
>>>> would be better to improve the interpretation of "revno:-$SMALLINT".
>>>
>>> Trying your patch on bzr.dev, it doesn't seem to make much difference to
>>> the results of 'time bzr version-info'. Maybe I'm doing something wrong.
>>> What speed-up are you seeing?
>>
>> Oh, I should have said: on emacs, 6s -> 0.2s.
> 
> Sorry for the fast follow-up: I think this is a branch5/branch6 thing.
> bzr.dev is branch5, the emacs import is branch6.
> 
> Cheers,
> mwh
> 
> 

Is bzr.dev still a branch5? I'm pretty sure we upgraded to branch6 a
long time ago. Branch 5 has some significant problems with scaling to
long linear histories. (As .bzr/branch/revision-history starts to get
*really* big.)

John
=:->

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFH2ZoeJdeBCYSNAAMRAjvTAJsED0GbUYm2yZaab2BjmysbtcdpXACfaqjH
p0KWZNC2DH3R6ndYGND39ts=
=X9od
-----END PGP SIGNATURE-----