--- "Stephen J. Turnbull" <[EMAIL PROTECTED]> escribi�:
> >>>>> "Antonio" == Antonio Regidor Garc�a <[EMAIL PROTECTED]> writes:
>
> Antonio> A general hunk patch can't replace a token replace patch
> Antonio> (and vice versa), but a concrete hunk patch can replace a
> Antonio> concrete token replace patch (and vice versa).
>
> I'm still confused by this. Is there a difference between a "concrete
> hunk patch" and a "concrete token replace patch"? A tree pair is a
> tree pair.
Yes, they are the same. This sentence was not much clear, sorry. A general hunk
patch is different
from a general token replace patch (they are different functions that transform
trees). On the
other hand, let's suppose that you have a concrete patch P Q and want to
interchange P and Q.
Let's say P = (R,S) and Q = (S,T). In David's notation, you obtain Q' P', where
Q' "does the same
change" than Q, and P' the same than P. In my description, P and Q commute if
there are two
general patches p and q so that P = (R,p(R)), Q=(S,q(S)), and P Q = (R,q(R))
(q(R),p(q(R))). You
have to _choose_ p and q, because many different general patches can generate P
and Q (and from
your example, general patches that make P and Q commute). So P can be for
example generated by a
general token replace patch or a general hunk patch. So, in this sense, you can
say that P is a
concrete hunk patch or a concrete token replace patch. But a general token
replace patch is never
the same than a general hunk patch. For commutativity, you can (in some cases)
choose a general
hunk patch or a general token replace patch to generate P, and thus interchange
P and Q in two
different ways. It's not a very good wording, anyway.
David says in the manual that he needs invertibility for theorem 2. And that he
uses this theorem
in the implementation, but don't know if it is necessary. So I will study the
implementation
before writing other theory.
Best regards,
Antonio Regidor Garc�a
______________________________________________
Renovamos el Correo Yahoo!
Nuevos servicios, m�s seguridad
http://correo.yahoo.es
_______________________________________________
darcs-users mailing list
[email protected]
http://www.abridgegame.org/mailman/listinfo/darcs-users
