back to yum
September 15, 2004
So I think all the respinning and signing and pain of the centos rebuild/respin are mostly over. This meant I got back to some work on yum tonight, after watching some west wing with the girl.
Yum stuff could be really good. I needed to work on the package verification on download because I’ve left out a couple of cases where the package downloads right, but is more or less the wrong package for completely ridiculous reasons. Anyway, writing this code to make the comparisons easier may have opened up an angle to work on local file installs from yum.
So maybe after I handle the goofball errors I can get to something like yum install /some/file/I/think/is/cool.rpm and have it just work.
Panu has been talking about some very interesting ideas on yum-devel list. Mostly about working on a python tool to use with the xml-metadata that let’s you do arbitrary queries against it. Sorta like how you can do queries via rpm from the command line against an rpmdb. If you could do the same general type of commands but to xml metadata repositories it could make searches and general reporting that much nicer. If anyone is interested in helping out look at the yum-devel list and see if there is code that needs to be written. Remember, the sooner this happens the sooner we can dramatically reduce the number of redundant copies of the rpm metadata that is kept in Fedora Core and other distributions.