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

Re: [PATCH] The blame speedup

From: Peter N. Lundblad <peter_at_famlundblad.se>
Date: 2004-05-31 18:25:52 CEST

On Mon, 31 May 2004, Greg Hudson wrote:

> On Mon, 2004-05-31 at 04:50, Peter N. Lundblad wrote:
> Perhaps it would be better to hand delta windows to the caller, as
> originally theorized. As has been pointed out, some day it may be
> possible to do something clever to calculate diffs from delta windows
> (although that sounds awfully hard and not very rewarding), and
> generally our RA interfaces match pretty closely the kind of data being
> passed over the wire, even over HTTP.
>
Yes, I'm coming to that conclusion too... This will lead to more delta
fiddeling in the ra_local case, but that's not what we are optimizing.
Don't know how many times one has to learn to prefer simplicity and clean
semantics to less important optimizations... I'll go for this solution
tonight.

> A helper function in libsvn_client/blame.c can translate the delta
> windows to temporary files.
>
Yes. I'll also try to merge the old and the new code, since we will have
to live with the old way of blaming during the 1.x cycle, I assume.

ONe little question. I'd like to give the client a svn_txdelta_stream_t*,
instead of having it provide a window handler and a baton. If think it
will be simpler to pull the delta when I need it instead of having to wait
for the data to be pushed from the RA layer. Are there any problems with
this that I don't see right now?

Thank you for being so patient with me:-)
//Peter

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Mon May 31 18:16:17 2004

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

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.