Re: [RFC] send-pack: allow skipping delta when sending pack

Jeff King [off-list ref] writes:

On Sat, May 20, 2006 at 11:17:42PM -0700, Junio C Hamano wrote:

quoted

base delta for that object to skip computation".  What you want
here is "if the object we are going to send is not a delta in
the source, and there are sufficient number of other objects the
object could have been deltified against, then it is very likely
that it was not worth deltifying when it was packed; so it is
probably not worth deltifying it now".

I think we can make a stronger statement in many cases: "if the object
we are going to send is not a delta in the source, and there are no
other objects it could be deltified against, then it is not worth
deltifying." That is, in the case that we just packed we KNOW that it's
not worth it, since we're not sending anything that isn't already
packed.

Careful.

We do not delta an otherwise perfectly deltifiable object if its
delta base happens to be at the depth edge in the original pack.
So no, we do _NOT_ know if it is not worth it merely from the
fact that it is not deltified in the existing pack.  And the
latter part of your test "there are no other objects it could be
deltified against" is either expensive (you have to try first to
see if that is the case to really see it) or stupid (you just
assume there is no suitable delta base without looking at other
objects like we currently do).

On Sun, May 21, 2006 at 01:31:22AM -0700, Junio C Hamano wrote:

We do not delta an otherwise perfectly deltifiable object if its
delta base happens to be at the depth edge in the original pack.
So no, we do _NOT_ know if it is not worth it merely from the
fact that it is not deltified in the existing pack.  And the

Yes, you're right. What do you suggest for this performance issue, then?
A manual no-delta trigger, or just going to get a cup of coffee while
pushing (my tests showed 5-6x slowdown from deltifying)?

-Peff

Jeff King [off-list ref] writes:

What do you suggest for this performance issue, then?
A manual no-delta trigger, or just going to get a cup of coffee while
pushing (my tests showed 5-6x slowdown from deltifying)?

In the short term, give --depth to send-pack, would be better
than nothing.

I agree we should be able to do better.  But let me grab some
sleep now first ;-).