[darcs-users] tag an ``excursion''?

[Jamie Webb]

On Tue, Jun 14, 2005 at 05:03:07PM +0200, Albert Reiner wrote:
> The real question is whether this inv(P)TPR will be equivalent to R as
> far as commutations are concerned, i.e., whether inv(P)TP = id is
> recognized and patches will commute as if inv(P)TP were not even
> there.

AFAICT, Darcs does not consider that inv(P)P = id, so, err, no.

This is probably because it doesn't quite work: consider what happens
if you freely commute across the rollback pair, and then decide to
unpull inv(P). Even as it stands there are bugs related to this. The
rollback design is AIUI considered irreparably flawed, and will
hopefully be replaced by something better at some point.

-- Jamie Webb