fast-import now rocks on brisbane-core

[Ian Clatworthy]

Ian Clatworthy wrote:

> On the bright side, importing into a gc-chk255 branch
> is now twice as fast as importing into a 1.9 branch.
> It's still a bit slower though that how fast fast-import
> was back in the old (pre VersionedFiles) days: it use
> to take a mere 1m50s IIRC (and the parsing layer is 30s
> or more quicker now).

Hmm - I forgot that 1m50s was on my desktop, not my laptop.
Testing on my desktop, importing that data set now takes
1m10s. Not bad for 6.1k revisions! So fast-import rocks on
brisbane-core+groupcompress! (Once we stop deserialising
inside chk_map.py so often, the time will be better still.)

I'm trying to import the linux kernel tonight. So far, it's
been running for 3.5 hours to import 40k revisions @ 180/min.
That's many times faster than some earlier figures I've seen.
(https://bugs.edge.launchpad.net/bzr-fastimport/+bug/259711)

More importantly, it means many large imports can now be
done overnight rather than take a week or more. It also means
it's feasible to convert to several candidate formats we're
considering to see how they stack up in our benchmarking.

Ian C.