From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.4) Gecko/20030703 Description of problem: Running e.g. psql from postgresql-7.3.4-2 will complain about relocation error in libpq.so.3 when used with krb5-libs from RH9 update. Version-Release number of selected component (if applicable): postgresql-libs-7.3.4-2 (krb5-libs-1.2.7-14) How reproducible: Always Steps to Reproduce: 1.Install RH9 with the upgrade krb5-libs-1.2.7-14 2.Install postgresql-7.3.4-2 and postgresql-libs-7.3.4-2 from FC1 3.Run psql Actual Results: uebn>> psql psql: relocation error: /usr/lib/libpq.so.3: undefined symbol: krb5_cc_get_principal Expected Results: Command prompt from psql. Additional info: I suppose postgresql-libs-7.3.4-2 should have a "Requires: krb5-libs >= 1.3.1-6". (Or at least "krb5-libs > 1.2.7-14".)
I'm wondering why you're using binaries that are compiled for FC1 on a RHL9 machine. The FC1 binaries are compiled specifically against FC1's library set, and they are not expected to work on any other platform besides FC1. We recently released postgresql-7.3.4-3.rhl9 RPMs as updates for RHL9, and you can retrieve them by running up2date. For RHL9 machines, please use these RPMs instead of the FC1 RPMs. Having said that, I think your suggestion about adding "Requires: krb5-libs >= 1.3.1-6" in the FC1 RPM is a valid point. It will be included in the next update for FC1.
To answer your wondering: The system isn't a pure RH9 system. I'm upgrading, but don't want to break everything at once, so I take it piecemal. Whenever I need to upgrade something, I take it from the most recent release. It is then of course reasonable for the new package to require the upgrade of other packages too to work properly. But if it does, it should say so in its requirements. Not that I expect everything to be perfect; I'm aware it is not easy to know all such details. But when I find it doesn't work I've considered it a bug and reported it. Like here! :-)
Ah, I understand your situation now. Thank you for clearing that up. And of course, you were perfectly right to have reported the bug. :) It is much appreciated.
This is now fixed in the Fedora Core 1 update postgresql-7.3.4-11.