> > Except that we don't want to remove prefixes in the case I gave below > > running old/new/old clients. I can try to code this up without sets, if > > you'd like. > > Ah, of course. No, sets are fine, though I'd ditch the for loop: > > self.attrs["prefix"] = "".join(pfx_set)
Thanks, this approach is cleaner. > > I'm a little lost, since I would have assumed that we would have written > > code to modify what we're looking for once prefix Z was introduced. I'm > > not adding new prefixes yet, so it seemed a bit overkill to modify this > > past explicitly ignoring unknown prefixes. > > > > I assumed that since the only prefixes that we know about in this case > > are C and V and we'll ignore all others, it's still sane to keep this > > code. Do you disagree? > > I'm coming around. I still think that the code ought to peel off one field > at a time from the left, making a decision after each field about what > needs to happen next, but I think you're making the right checks in the > end. I'm not opposed to doing things this way. I was under the impression that this is what we're doing, or at least attempting to do now. I've posted a newer webrev here: http://cr.opensolaris.org/~johansen/pkg-unknown/ Let me know if this looks like it does a better job of checking the fields that need to be checked. -j _______________________________________________ pkg-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
