Daniel Berlin wrote:
> On 3/29/07, Malcolm Rowe <malcolm-svn-dev@farside.org.uk> wrote:
>> I'm not sure whether this is the cause, but be aware that we compute
>> the deltas using a window of 100K that moves (in 100K steps) through the
>> reference and version files, so we can't take advantage of inter-window
>> deltas. We do this to bound the memory usage of the compressor and
>> decompressor, and it seems to be one of the most significant factors in
>> our compression performance, especially if we ever want to start using
>> some of the (theoretically) better delta algorithms.
>
> Ya, we need to up the window size (to about a meg or two), but it's
> hardcoded in some places that make it a not backwards compatible
> change with more work.
Yep, that's definitely the, problem. To be fair, Subversion actually
out-performs xdelta3 with a window that small, my a margin of about 5M.
On the xdelta.org site, they imply that large files use a window of
100Mb, which roughly adds up (my original tests used a total of 6
windows for the whole comparison, while the test I did today used a
total of around 450 windows).
Hmm, I wish I had time to pick up the Subversion source and see if I can
fix this, but I have too many other priorities. Maybe in a couple
months I'll see what I can do, if somebody else isn't already on it.
-Rick-
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Thu Mar 29 23:46:12 2007