On Sat, Sep 07, 2002 at 09:08:26AM -0400, Robert G. Brown wrote: > On Fri, 6 Sep 2002, Connie Sieh wrote: > > > As long as we are talking about the first pass to get the headers. I have > > seen it just stop alot. I installed yum on my system at home. It has a > > cable modem. Yum would download a few headers and then stop. I would > > restart it and it would get a few more. I resorted to getting them via > > ncftp. > > > > I was thinking that we might have to package up the headers as a rpm and > > install it when we install yum. This will thus seed the cache and yum > > will then only have to get new headers. > > Absolutely fabulous idea -- I was thinking something similar but didn't > think of it being a "pre" rpm, just a tarball. I also run installs from > home over DSL and the system I borked was a DSL install. It is a heck > of a lot easier (and faster) to install just one big file than a lot of > little files and just do the updates. > > The rpm (or tarball) could probably be generated at the same time as the > headers themselves, by the same script. To take this one step further... yum could detect whether it should get them all separtely, or just download the tarball. That way, after a fresh install or a "clean headers" it would just say "oh... I don't have _any_ headers. Let's just get the tarball". This wouldn't be hard, would it, Seth? It seems that a tarball is probably more logical. What would the header-rpm name be? What would the version be? What about header-rpms from two different repositories? rpm comes with too much baggage for this application IMHO. -Michael -- Michael Stenner Office Phone: 919-660-2513 Duke University, Dept. of Physics mstenner@xxxxxxxxxxxx Box 90305, Durham N.C. 27708-0305
