slower performance of log and annotate in 1.6?

[John Arbash Meinel]

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

Robert Collins wrote:
> On Mon, 2008-08-11 at 20:52 -0500, John Arbash Meinel wrote:
> 
> 
>> As I see it, the problem is that to find the per-file graph, you have to
>> indirect through the inventory, as we no longer have a way to say "give
>> me the graph of *this* file_id" because we don't have VersionedFile
>> (singular). It was bad to use "VF.versions()" because it had to load the
>> whole index, but we don't have an *entry-point* into the partial graph
>> anymore.
> 
> Strictly speaking we do - we have iter_entries_prefix, and could use
> that.
> 
> I'd like to know how it performs under a btree index though.
> 
> -Rob

We have that at the layer of index, not at the layer of VersionedFiles.
It was the old KnitVersionedFile.versions() that thunked over to
iter_entries_prefix (which I think triggered a _buffer_all).

I'll try to benchmark a btree version tomorrow.

John
=:->

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

iEYEARECAAYFAkihASoACgkQJdeBCYSNAAMlcwCgqHoqCwA2gVqxG6GQvr7AhV+0
vmoAnA+Q0+fxv2XnB9tNJvHa+BmdBgcU
=pZ4D
-----END PGP SIGNATURE-----