Re: isn't variance adjusted patching horribly dangerous?

I feel like this whole discussion of whether variance-adjusted
patching is "dangerous" or not is something of a useless tangent right
now.

The paper itself points out that the algorithm may be overeager for
some use cases, and recommends attaching "selectivity knobs" to the
various steps in the algorithm, to avoid this overeagerness when
appropriate. If you're versioning wedding invitations, then turn the
knobs down; for code, turn them up.

As to where the knobs' defaults should be, that's a question that can
be settled by experience and/or debate when the feature is actually
available. There are a lot more important things to be worrying about
than this right now, even just in the area of improved merge support.

Just my $0.02,
-Karl

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Thu Apr 10 18:56:02 2003