[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: [ITP/mostly packaged] hftpd



Adrian Bridgett <adrian.bridgett@zetnet.co.uk> writes:
> On Thu, May 20, 1999 at 01:14:49PM -0700, Chris Waters wrote:
> > Adrian Bridgett <adrian.bridgett@zetnet.co.uk> writes:
> > 
> > > On Tue, May 18, 1999 at 05:09:27PM -0400, Michael Stone wrote:
> > > [snip]
> > > > > We really should have a policy for things like this. How about adding
> > > > > another Provides: to kernel images (built by the excellent make-kpkg):
> > 
> > > > Because too many people don't use debian kernel images.
> > 
> > > How about only making dependant packages "suggest" the appropriate kernel
> > > version?  That gets around this problem.
> > 
> > Or make it conflict with inappropriate versions?  (Actually, this
> > would work better if kernel-image-2.0.xx provided the virtual package
> > kernel-image-2.0, etc., but whatever.)
> 
> No conflicts - you can install as many kernels as you like - you could have
> 2.0 and 2.2 installed and then boot between them :-)

The start of this thread is vague in my memory now, so apologies if I'm
off target here, but I would object to any sort of enforced kernel version
dependancy on a package.

Just because I have the 'correct' kernel version for a package installed
is no guarantee that I am actually running it, so a dependancy doesn't
really achieve much.

I tend to use make-kpkg if I'm building a kernel to use on a different
machine than the one I'm building it on, but otherwise I usually dont.
Kernel packages are handy if you wish to intstall the same kernel image
on multiple machines, but since all my machines have different hardware
in them I usually do a custom build for each.  For machines I keep
kernel source on, I'm usually happy to build a zImage without the
overhead of packaging it and its modules too.. (yes this is slight,
yes it's also personal preference..)

My view is that if a package requires a certain kernel version, then
say so clearly in it's description.  If people choose to ignore that,
well.. caveat emptor...

best,
Ron.


Reply to: