On Wed, 19 Nov 2003, Jeff Johnson wrote: > On Wed, Nov 19, 2003 at 01:16:08PM -0500, James Olin Oden wrote: > > On Wed, 19 Nov 2003, Jeff Johnson wrote: > > > > > > I've updated RH9 -> FC1 and have a number of troubles: cvs doesn't > > > > work, for example. I suppose that there is some relationship with > > > > these warnings... > > > > > > > > > > You've specified glibc multiple times on the command line, probably > > > because there are both i386 and i686 > > > packages and you are using a wild card. There are important differences > > > between i386 and i686 glibc, only > > > the i686 package has NPTL. > > > > > Hey Jeff, > > > > I get this exact same warning when using rpm with my autorollback patch. > > I think your explanation explains why. What might I do to get the message > > to go away? Just as a brief re-cap of what the patch does, it builds a > > rollback transaction as it is running a transaction. Each time something > > sucessfully installs/erases that package (or header) gets added to the > > rollback transaction. If I have no failures I discard the rollback > > transaction, if something fails I run it. Anyway, so I have two > > transactions instatiated at the same time in librpm, and I suspect from > > what you said this is where the funny little warnings are coming from. > > So again, what might I do (or what code might I read (-;) to aleviate > > these little warning messages? > > > > Don't add more than one identically named packages to the same transaction > is the answer. Makes no sense imho. More objectively, results are undefined > until what is supposed to happen is concretely and completely defined. > That is my point, I am not adding more than one identically named package (or at least I don't think I am) to the same transaction. There is one running transaction, and as it is running I am building another transaction (which is the reverse of the transaction running). It seems like even though the two transactions are seperate, and neither one of them has duplicate NEVRA's within themselves I am still getting this message. I could be wrong so I will double check. Cheers...james > What is defined now is that a warning message is spewed, and the newer > EVR (note no explict A, but package Elf32/Elf64 color is checked) replaces > the older. Still, developer brain fart imho. > > E == epoch, A == arch, etc as usual. > > 73 de Jeff > > _______________________________________________ Rpm-list mailing list Rpm-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/rpm-list
