On Fri, Apr 30, 2004 at 08:19:42AM +1200, Michael Honeyfield wrote: > Hello. > > I am working with RPMv3.0.x at this sage. I would like to see RPMv4.x.x put in use. However, the person in charge of such choices will not accept my answer of "Because we should..." > > So.... > > Can people here tell me why, both technically and for an end user, we should move from RPMv3.0.x (which is working fine) to the current version? > The primary reason to move from rpm-3.0.x is to avoid catastrophic database loss from "Free list corrupt." The installed package headers are saved with a clever-but-imperfect first fit scheme in the file /var/lib/rpm/packages.rpm. In rpm-4.0.x, the scheme is replaced with a Berkeley DB hash, so any damage has a prayer of being repaired using utilities provided with Sleepycat Berkeley DB. Re-install is the only answer I have for "Freelist corrupt." Note that rpm-4.0.4 can often reconnect the chain of headers. OTOH, I have exactly zero interest in backporting the fix, rpm-3.0.x is dead code, and support for the first-fit rpmdb scheme has been entirely removed in rpm-4.1. > Just to help narrow things down.... > > We have no interest in the python API bindings, well, I do, but I am not making the choices. > > We have no interest in writing specs in xml. > > So other than those two features... what else it there? > There's a slew of bug fixes, the most important (imho) of which is the "Freelist corrupt fix." OTOH, there ain't nothing wrong with rpm-3.0.x either. If your machine "just works" with rpm-3.0.x, well, live with that. If you have any desire for support, or upgrades, I'd suggest getting rid of the first-fit scheme in favor of Berkeley DB would permit support/upgrades. 73 de Jeff -- Jeff Johnson ARS N3NPQ jbj@xxxxxxxxxx (jbj@xxxxxxx) Chapel Hill, NC _______________________________________________ Rpm-list mailing list Rpm-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/rpm-list
