On Wed, Aug 27, 2008 at 01:30:38PM -0700, Jason Dagit wrote: > > I understand that, but your changes are wrong with respect to the > > code, and that's worse. We may as well make Repository.Internal > > right, since we can't make the commands right. > > I disagree that your changes are any more correct.
Well that's unfortunate. > > And the truth is that the tentative state *does* exist, and we > > *should* document it, and we should document it correctly. Certain > > functions (e.g. those I changed back) can be run after a previous > > function has changed the tentative state, and this needs to be > > documented. Alas, we have no tracking of changes of the tentative > > state, but that just means that we gain nothing by converting all the > > commands to use type witnesses. It doesn't mean that we can't gain > > anything in converting Repository--when we convert Repository, we gain > > documentation, and I want it to be correct. > > Near as I can tell, tentative state t is not well-defined but tr and > tu exist and are well-defined. No, t is well-defined, is the same thing as tr, and tu doesn't exist. David _______________________________________________ darcs-users mailing list [email protected] http://lists.osuosl.org/mailman/listinfo/darcs-users
