o 14/5/2004 YC

- creation of installation program should be parameterized with version
platform
- installation program => problems with absolute vs relative path

o 7/12/2003 MMM

- I think that there are too many targets in the Makefiles.
  For instance, at the root of the forest, there are 6
  targets for cleaning. Nobody is going to remember them.
  I think that "cleandistrib" (for all the builds outside
  of the forest) "cleansource" or "cleansources" 
  (for all the builds  inside) 
  and "clean" (for doing both) would be enough


- The documentation of some makefiles are not consistent
  with the actual targets. For "lib/aldor/Makefile", I
  just removed these comments.

- The style of the different makefiles is not consistent.
  Even worth: some makefiles that were writen by different
  authors with different styles contain DEAD CODE. I think
  this is dangerous. (That's a way to do "rm -fr *" without
  meaning it.)

- "lib/algebra/Makefile.generic" seems to be the Makefile
  of the distributed "libalgebra". It does not include a
  target for running the tests (see how the former "Makefile.algebra")

- Concerning the targets of the Makefiles for "algebra"
  and "libaldor". I think it is better to turn every target
  into a script call. This way the Makefiles for the users
  (which have generally less targets) and the Makefiles for
  the developers can share the same code.

- When doing "make generic" at the root of the forest.
  some .al archives stay in the lib directory of their
  source tree (aldor, algebra) other are moved automatically
  to the distribution tree (axllib).

       ==> This needs to be fixed


