Re: Easy Installs

["Matthew C. Weigel" <weigel+@pitt.edu>]

On Thu, 1 Jul 1999, Julian R Fondren wrote:

> and of course, dpkg and rpm both use huge databases and have funky
> dependency ideas like: "If my database says you don't have it - you don't
> have it!" and "If my database says you have it - you have it!" whereas the

Which is why dpkg uses a surprisingly readable flat text file to store
information on what is in the system.  If you add external software
manually, you can simply add ``appending this piece of software to the dpkg
database'' to the list of things that you manually do.  It can even be stuck
into the Makefile.

dpkg was designed, strangely enough, to be a useful tool for real situations
-- situations where maybe you didn't install everything by Debians to begin
with, or maybe you installed dpkg itself after the fact.  dpkg is still
useful for software that *does* come in a .deb file, and doing everything it
does manually is actually pretty good discipline for a large system.

rpm may or may not do everything I've mentioned above for dpkg, I don't
know.  I *do* know I've had relatively negative experiences with rpm getting
confused about installed packages, and not do too well with several levels
of dependancy.

And of course, this is hardly an argument for OpenBSD to go to dpkg. 
Although free software, dpkg is hardly used anywhere but Linux, and a .deb
is simply a tarball with a specific directory structure.  So the only real
benefit is enforcing some sysadmin discipline which should be there anyways.

 Matthew Weigel                                       Programmer/Sysadmin
  weigel+@pitt.edu                             Operating Systems Advocate
                         http://www.pitt.edu/~weigel