Re: mol comments
Jonas Smedegaard wrote:
>
> On Thu, 26 Apr 2001, Michel Dänzer wrote:
>
> > Jonas Smedegaard wrote:
> >
> > > I was hoping you would have made the module-source hooking into
> > > kernel-package to make building it semi-automatic. I haven't figured out
> > > yet just how to do it myself, but if not anything else you could maybe
> > > simply throw it into the TODO :-)
> >
> > Please don't depend on kernel-package though. It can easily pick up the
> > correct tree for the running kernel if it's 2.4.x (and even current
> > 2.2.x?) using /lib/modules/`uname -r`/build .
>
> I am not sure I understand you correctly (or you me?).
>
> What I mean is set up the mol-modules-source package similar to
> alsa-source and device3dfx-source (which is a small package that I have
> never compiled but might be useful as reference for making it work with
> kernel-package.
>
> At least sheep_net requires kernel-source (or kernel-headers) to build,
> not only a running kernel.
/lib/modules/`uname -r`/build is a link to the tree the running kernel was
built with. (Provided that make modules_install was done). That should be
enough to build a kernel module.
> Maybe putting sheep_net in a package for itself (that would make it
> available for other architectures as well) and include the source and a
> compile-script in mol-modules for the rest (which I believe I succeeded
> compiling without kernel-headers available).
Sounds good.
> The current situation is bad: mol-modules-source creates a kernel-modules
> .deb of the _same_ version as the official .deb. The naming scheme of
> kernel-package (and kernellab) avoids that in most situations.
I have to admit I don't really care about this as I don't use kernel nor
module packages. All I'm asking for is not to impose a dependency on either.
--
Earthling Michel Dänzer (MrCooper) \ Debian GNU/Linux (powerpc) developer
CS student, Free Software enthusiast \ XFree86 and DRI project member
Reply to: