Shawn, > > Would you clarify how the Atom entry's ID is going to be formed? > > If you're suggesting that we simply use the FMRI, this fails the test > > for an, "ID [that] must be globally unique, across all Atom feeds, > > everywhere, for all time." > > > > In the unlikely event that two depots publish the exact same package at > > the same version at the same time, you'll have a namespace collision. > > Actually, I'm expecting them to. > > That's the problem I'm trying to figure out. If the same package is > republished at a different repository, it should still have the same id > shouldn't it? > > I'm intentionally trying to ensure that it happens. In fact, one of the > topics brought up on that "diveinto" article is ensuring the rss entries > that are re-published at an aggregator somewhere else still have the > same id.
I think we're getting package namespace constraints intermingled with our discussion of Atom entry IDs. If two depots publish a package with the same name, Stephen's namespace rules state that these packages are expected to deliver equivalent functionality. However, equivalent functionality isn't the same thing as the two packages being identical. They may contain different files and actions. I think that we have to assume that each package published by a depot is globally unique, even though it may be functionally equivalent to something else. This, I presume, is why we'd need to have a unique entry ID for each package. -j _______________________________________________ pkg-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
