Re: LILO 21.6-2
>>"Peter" == Peter Makholm <peter@makholm.net> writes:
Peter> Manoj Srivastava <srivasta@debian.org> writes:
>> At the moment, if there is no lilo.conf, the kernel-image
>> postinst creates a functioanl lilo.conf that takes into account
Peter> Is this wise?
You are asking my opinion? Absolutely.
Peter> I assume that in a perfect world the lilo package would be
Peter> better to configure itself than some "random" other
Peter> package. The lilo package would have more knowledge of which
Peter> special things to consider, eventually after consulting the
Peter> user.
That does not yet happen to be the case.
Additionally, this functionality has been in the kernel image
package for about 4 years now; and lilo.conf was never a conffile.
Additionally, creating a lilo.conf is done with full admin
approval -- and, IMHO, does a reasonable job of the lilo.conf
file. In absennce of any capability in the lilo package this was
required to have a system bootable after installing a kernel
image. Now that lilo may be doing the job on its own, the need for a
``random'' third package to handle what should always have been
lilo's job is reduced.
I suggest that the lilo package create a /usr/sbin/liloconfig,
or some such, which can be invoked at the admins behest and
convenience.
Peter> Consider this: Kernel-image gets installed and finds no lilo.conf and
Peter> makes a minimal but functional lilo.conf. Then lilo gets installed, it
Peter> finds a lilo.conf and decides not to touch it (what a nice
Peter> package). What the user now doesn't know is that the lilo packages
Peter> configuration of lilo.conf would have given him the oputunity to use a
Peter> password on lilo or some other fancy feature.
Nice hypothesis -- and when it becomes close to reality
kernel-package shall be changed. All I was commenting was that the
lilo config needs to be improved to subsume the capability of the
kernel image postinst (surely not too much to ask for).
Peter> Nobody should touch the configuration files belonging to lilo but the
Peter> local sysadmin and the lilo package itself. And the lilo package
Peter> should rather be a little less fancy in it's install than destroying
Peter> existing configurations.
Were lilo.conf a conffile you would be correct. But it has not
been so until fairly recently; and, in any case, no lilo.conf was not
better than a minimal one. I have never said that once we have a
reasonably competent lilo config process I would not change
kernel-package.
Incidentally, an argument could be made that the local admin
created the file, using the postinst as an intelligent language
sensitive editor rather than vi.
manoj
--
Magnocartic, adj.: Any automobile that, when left unattended,
attracts shopping carts. Sniglets, "Rich Hall & Friends"
Manoj Srivastava <srivasta@debian.org> <http://www.debian.org/%7Esrivasta/>
1024R/C7261095 print CB D9 F4 12 68 07 E4 05 CC 2D 27 12 1D F5 E8 6E
1024D/BF24424C print 4966 F272 D093 B493 410B 924B 21BA DABB BF24 424C
Reply to: