On Wed, 2005-07-13 at 22:44 -0400, Carl Nygard wrote: > I've seen mention various places (but all available docs are out of > date) that pieces of anaconda placed in a RHUpdate dir located somewhere > strategic will get included by the buildinstall process as part of the > new anaconda. > > Would it be possible to add the additional installclass/newclass.py file > (sans newclass.pyc) and have it just work? I looked at the code, and > the installclass/*.py are loaded via a directory scan, so it seems > possible. Anyone have more specific clues? Using RHupdates here really isn't probably what you want. In fact, unless you're doing testing while doing development, an updates.img is almost always a better idea. But in this case, you probably want to have product[1]/base/product.img instead. This should be an ext2 formatted disk image with an installclasses directory in the root of it. Then, place all of the installclasses you want to be available at install time in that directory. For more details on the way it works, see installclass.py:availableClasses(). Cheers, Jeremy [1] Where "product" is Fedora if you're using Fedora Core, RedHat for Red Hat Enterprise Linux, etc.
