On Nov 27, 2007 10:19 AM, Todd Zullinger <tmz@xxxxxxxxx> wrote: > Anyone that's parsing rpm output in a script should have been using > the --qf option a long time ago to ensure that they get the output > they want, really. The --qf option even for uninstall package files sitting on a filesystem which may or may not be mounted on an operating system with rpm available but is accessible via the network for rpm using system to pull from? We can should ourselves until we are blue in the face and tsk tsk people for not using the qf mechanism appropriately with scripted rpm queries... but we'll still need to make a polite effort to re-educate the people who are relying on the default structure of string. I'm not saying this as an argument against doing it. I'm saying when you do it, it will be relatively more painful to add a new epoch field internal to the string than tacking on information at the end of the querystring AND the package filename. We should make a vocal preemptive effort to get the word out before this lands in a release to take the edge off the disgruntled bloodlust for making a change. There is discussion in this thread about changing the filename as well as the rpmdb querystring to be self-consistent in terms of displaying the epoch. Any scripting against package filenames will be disrupted by a new epoch field and I think that deserves some effort to raise awareness sooner rather than later. -jef -- fedora-devel-list mailing list fedora-devel-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/fedora-devel-list
