Subject: Re: Wher do binary packages go.
To: Kevin Sullivan <ksulliva@kludge.psc.edu>
From: VaX#n8 <vax@ccwf.cc.utexas.edu>
List: current-users
Date: 03/15/1995 03:47:17
Disclaimer: I use the term package to mean a big chunk of source code.
I considered the whole binary/package/directory thing a while back, and
while I originally started with source packages in an obscure directory,
and the links in /usr/local/bin to the packages, I figured this was a waste
of CPU time to track the symlink each time.  Better, I thought, to put
the bins in the bin directory and simply link from the package dir, but this
had it's problems.

I came to two conclusions:
1) I was trying to use one tree structure for two things.  That is, I was
trying to both document what was in a package, and I was trying to isolate
based on file types.  (And some OS's try to distinguish traffic/persistence
as well -- as in the /var, /tmp thing)  This can be done, but it should be
done more carefully and probably done by revolution than evolution (which
seems to be contrary to Unix history).
2) If makefiles/scripts were smarter, you could "uninstall" packages easily.
And keep track of what files are in what package.  I can't count the number
of times I looked in my bin directory and thought, "what is THAT for?".
Now that I think about it, it might not be the right question for Makefiles;
they might be more suited for "what files are in this package, and where
are they installed?"  Basically, makefiles and scripts could take a great
burden of documentation off the file system tree.  Haven't looked into
binary packages yet, since I'm a CVS/source code kinda guy.
-- 
VaX#n8 (vak-sa-nate) - n, CS senior++ and Unix junkie - vax@ccwf.cc.utexas.edu
Vax es mi nombre, queso es mi juego.                  - PGP key on request