On Mon, Jul 16, 2018 at 1:21 PM, Penney, Don <Don.Penney@windriver.com> wrote:
I started writing a script to pull together a list of the dependencies based on Requires and BuildRequires. As I understand it, we need to add all such dependencies, recursively, to these .lst files. Using the StarlingX.repo config file as an arg to repoquery, I get some packages overlapping with rpms_from_centos_3rd_parties.lst, however, so I’m hoping there’s some clear direction with these files. My script is finding hundreds of rpms to add to rpms_from_centos_repo.lst with its calls to repoquery to resolve the dependencies, so either I’ve got lots of overlap and many of these need to go to the 3rdparty file, my repoquery approach for automating this is completely wrong, or these new packages are using a completely different set of dependencies for the builds than anything else we’ve got.
Thank you! I haven't had time to look at that yet but have been hoping that we could generate even more of this stuff from the other parts of the repo. As for the volume of dependencies, it's been a while since I did something like that with RHEL, but it has always had copious amounts of dependencies even for "small" installations. (Does DBus still require X on a server? I don't think so, but it once did...)
I’ve attached the first draft of my script for anyone that’s curious.
The attachment was bounced by Intel's MTA because of the .sh attachment so @intel.com addresses missed this. I'd go ahead and propose a WIP review[0] to stx-tools and iterate on it there. dt [0] In case not everyone knows the common practices in Gerrit, WIP reviews usually begin the commit message with 'WIP' or the author/committer can set a Workflow -1 in the Reply dialog. Or both. Often this means it is not ready for detailed review or it could be a way to solicit feedback during development, it is convenient to mention the reason for WIP in the commit message. -- Dean Troyer dtroyer@gmail.com