[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: faster client pre-1.0: neon prefetching, multithreading

From: Branko Čibej <brane_at_xbc.nu>
Date: 2003-02-11 20:53:25 CET

Greg Hudson wrote:

>On Tue, 2003-02-11 at 14:31, Branko Čibej wrote:
>>Oh, but I think mod_deflate can do better than vdelta as far as
>>compression is concerned (well, slightly better)
>If we ever dusted off Dan Berlin's svndiff1 stuff, I think we would do
>about as well as gzip spacewise. More importantly, though, would using
>zlib actually be any faster than using self-compressed vdeltas, if the
>goal is to speed things up?

The comparison was "diff -e | gzip", which was about the same size as
vdelta. I believe that for compression only, zlib is still significantly
better. I don't know about which is faster, but the real question is
whether the difference in time balances the difference in network
transfer size. I suspect it does, but who knows? We need performance
measurements, obviously.

>>Yes, and a smart compiler probably already does that.
>svn_txdelta__insert_op is defifined in a separate source file. So
>nothing using the normal Unix build process will do that, because the
>information isn't available at the right time. (Unless you put off all
>compilation until the linking phase, anyway.)
Ah, right. And we don't want to move that function into another file.
Well, the 1% is probably reallocation costs, anyway.

Brane Čibej   <brane_at_xbc.nu>   http://www.xbc.nu/brane/
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue Feb 11 20:54:11 2003

This is an archived mail posted to the Subversion Dev mailing list.