On Wed, 2007-06-27 at 07:54 -0400, Russell Harrison wrote:
On 6/25/07, Christopher Blizzard <blizzard(a)redhat.com> wrote:
I actually think that we need to capture a small bit of that
information. i.e. kernel, hal, pm-utils and whatnot. But
catching
everything? Probably not really needed for hardware
compat. (Maybe
other things like NM, bluez-utils, etc.)
Just trying to keep the scope pretty small. :)
Understood, it just seems to be about the same amount of codding
work to record everything rather than filtering stuff out. Now when
you start talking about storage required. . . that is a different
story. I think the reason I said it should be on the road map is in
the future smolt can help with software compatibility as well. Again
not a high priority given the current scope of smolt, just one for the
list.
Right.
What may be very valuable from a hardware compat standpoint is a
record of packages that aren't in the fedora repos, especially kernel
modules. We know people are going to be installing the nvidia and
fglrx drivers, but what else are they installing because the stock
kernel doesn't have drivers that do what they want.
I actually think that "loaded kernel modules" is one of the things we
want to capture for hardware compatibility questions. You could ask
questions like "show me all the common modules for laptops that don't
suspend" and you might learn a lot from the query about which modules
may or may not be bad. At least it's a decent set of data to query.
--Chris