[MERGE] Remove knit caches

[Robert Collins]

On Sun, 2006-06-25 at 19:21 -0400, Aaron Bentley wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Hi all,
> 
> Here's a patch that eliminates knit caches.  Theoretically, they could
> speed up repeated reads or requests for multiple versions of a file, but
> those operations are already fast, now that we have operations like
> get_texts.
> 
> Knit caches don't provide a significant performance benefit for commit
> operations-- the commit benchmark shows a 1% performance degradation
> that may just be noise.
> 
> The log_verbose benchmark shows a 1% performance degradation, which
> again is probably noise.
> 
> There aren't many operations that request multiple versions of a file.
> Annotate gets multiple revisions.  It shows a 12% slowdown, but it
> hasn't been optimized by using Repository.get_revisions
> 
> Even if the performance gains were more compelling, this would still be
> a choice between performance and correctness, because the caches consume
> an unlimited amount of memory.
> 
> Therefore, I propose that we get rid of the knit caches.  It seems to be
> best to do them at a higher level, instead.

I put them in for conversions from weaves - multiple inserts of non-knit
data.

The win there was to take some operations from hours down to minutes,
particularly for inventory insertions.

I'm happy for them to be taken out, as I was never particularly happy
about them being in, but the conversion path should be kept fast.

-Rob.


-- 
GPG key available at: <http://www.robertcollins.net/keys.txt>.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 191 bytes
Desc: This is a digitally signed message part
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20060626/c88707f0/attachment.pgp