Re: Package priorities and dependencies.
On Sun, 15 Jun 1997, Thomas Koenig wrote:
> Dale Scheetz wrote:
>
> >Two packages in the list of "important" refused to install because they
> >declared (correctly) their dependence upon packages of lower priority.
> >
> > at depends on libelf0 priority: optional
>
> This dependency isn't needed... hmm...
>
> For some reason, the configure script created by autoconf always
> looks for -lelf and, if it can find it, adds it to the list of
> searched libraries.
>
Probably because the author considered that this would make a better at.
> First, why does my autoconf - generated configure search for -lelf and
> then always adds it? IMHO, this is a bug in autoconf; I never asked it
> to do that.
It gets configured somewhere. Autoconf didn't come up with that on its
own.
>
> Second, this is a bug in at; I'll uninstall the libelf binary from
> the system I compile at on, so this doesn't happen again. That will
> be fixed in the next release, although it's far from being a critical
> bug :-)
While I'm sure that removing libelf from the development environment will
do what you suggest, I'm not sure that you either need or want to do that.
The actually dependency is determined by dpkg-shlibdeps found in the rules
file shortly before the package build. This actually interrogates the
program file, given as a parameter, using ldd to determine which libraries
it has been linked against the executable. These get added to the
dependency field by dpkg-gencontrol just before the package is built.
This means that the executable was, in fact, linked against libelf0. This
implies some functionality, derived from those linked routines that the
author, at least, thought of as beneficial. It is also entirely possible
that removing them from the output of autoconf will fail to result in a
viable executable (unresolved etc...) although many packages are designed
to be built under differing conditions. This makes it possible that at
will build without it just fine, but with reduced capability or speed of
operation.
Even if libelf0 only gives minor improvements in the package, the fact
that the author intended to take advantage of that librarys capabilities
where possible makes it desirable to keep it as a dependency.
>
> >This tells me that libelf0 and libg++27 should have their priority field
> >changed to "important".
>
> In principle, you're right :-)
Thanks for supporting the principle ;-)
I guess I was pushing for more practical application of the principle in
these two cases.
Waiting is,
Dwarf
--
_-_-_-_-_-_- _-_-_-_-_-_-_-
aka Dale Scheetz Phone: 1 (904) 656-9769
Flexible Software 11000 McCrackin Road
e-mail: dwarf@polaris.net Tallahassee, FL 32308
_-_-_-_-_-_- If you don't see what you want, just ask _-_-_-_-_-_-_-
--
TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to
debian-devel-request@lists.debian.org .
Trouble? e-mail to templin@bucknell.edu .
Reply to: