On Sun, 2006-07-09 at 20:15 +0200, Hans de Goede wrote: > I was thinking about how right you are on this and then I had a very > good idea, someone (with more experience with this then me and buildsys > / cvs server access) should write a script which periodically > 1) runs rpmlint on all RPMs > 2) compares the output of this script with the following file in CVS: > <packagename>/<branch>/rpmlint.ignore > and removes all lines from the output which match with lines in this > file. > 3) mailes the maintainer of the checked package if there is any ouput > left after removing the ignored lines > > I'm really enthousiastic about this idea, it would be great for QA and > should be easy to implement! My only worry is the possible resource > drain on the system doing the checking. I've been eyeballing this every now and then too but haven't got around to doing anything about it. Debian has something very similar if not exactly the above in place already plus more, perhaps something could be borrowed from there: http://lintian.debian.org/ BTW, it should be trivial to implement the rpmlint.ignore part, just make it a rpmlint config file, add addFilter()'s as necessary (see /usr/share/rpmlint/config for examples) and invoke rpmlint with "-f /path/to/<packagename>/<branch>/rpmlint.ignore". Folks should keep a pretty close eye on commits to those files though. > Also I think that asap the package submission prococedure should be > modified to make it mandatory for submitters to run rpmlint themselves > and include the output in the initial review submission including > motivations for the ignoring of each warning / error given by rpmlint +1. It should be also allowed and encouraged to just submit and ask for help if one doesn't know a good way to handle error/warning X instead of just desperately trying to get a clean output no matter what before the submission. -- fedora-extras-list mailing list fedora-extras-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/fedora-extras-list
