[RFC][DRAFT] Updated guide for upgrading to rich-roots [and the 2.0 beta formats]

[John Arbash Meinel]

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

Alexander Belchenko wrote:
> Alexander Belchenko пишет:
>> I've just tested upgrade of qbzr trunk (revno.769) to 2a format.
>> Everything seems went OK, repo shrinks 4.5M -> 1.5M.
>> But after upgrade I have 4 or 5 packs in the repo, and only when I've
>> repacked it I can see major speed-up in operations.
>> IIRC, this problem about needing pack after upgrade is known,
>> but I'm mentioning it anyway.
>>
>> IIUC, 2a is not the latest one bbc format?
>> I will wait for BIG announcement when bzr will be ready
>> for final upgrade.
> 
> 
> But something is bad with 2a. Branching without shared repo is slower
> with 2a.
> 
> timeit bzr branch trunk-1.9 test
> time: 6.424
> 
> timeit bzr branch trunk-2a test2
> time: 9.202
> 
> And we have fairy small history :-(

It is a known regression. Most of it is because we just have more small
pages to copy, some of it because analyzing all of those small pages is
slower (than passing a regex over XML delta lines). I wouldn't say I've
really seen the regression with smaller histories as strongly as you see
them.

It could also be different behaviours of the StreamSource fetch path,
versus the old Packer one.

It was something I tried to address a couple of times, but I was only
seeing it with Launchpad sized trees. I should poke again with something
the size of qbzr.

John
=:->

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

iEYEARECAAYFAkoyVJEACgkQJdeBCYSNAAOCdACghDcSbEoYX7HyKvJoUCWwPcT6
IJAAn0bn0OoNKrjF+urc+mMPDueEnlnB
=wrUJ
-----END PGP SIGNATURE-----