[MITgcm-devel] Genmake2 in good form

[Ed Hill]

On Tue, 2004-07-06 at 13:57, Chris Hill wrote:
>  "genmake2 rocks" t-shirts coming soon.


In all honesty, I think genmake2 in its current form deserves an "ok" at
best.

The GNU auto-tools (automake, autoconf, libtool, etc.) have come a long
way in recent years and they include much better versions of many of the
tests ("does ___ exist on this system?") that have been recently added
to genmake2.  I think the best long-term strategy is to take the
essential genmake2 functionality (e.g. the "packages" bits) and turn
them into functions running inside the autoconf framework.  And then use
autoconf for everything else.

For instance, NetCDF is a mix of F77, F90, C, and C++ and they've
managed to do a fairly good job integrating it all and getting it to
build with many different compilers on many different platforms using
autoconf.

And on the bright side, both genmake2 and the autoconf configs are
written in (mostly) Bourne shell so it won't be too hard to integrate
them.

Ed

ps - Read the following quote recently and wanted to pass it on:

  "Those who do not understand autoconf will be forced to 
   re-implement it, badly."


-- 
Edward H. Hill III, PhD
office:  MIT Dept. of EAPS;  Rm 54-1424;  77 Massachusetts Ave.
             Cambridge, MA 02139-4307
emails:  eh3 at mit.edu                ed at eh3.com
URLs:    http://web.mit.edu/eh3/    http://eh3.com/
phone:   617-253-0098
fax:     617-253-4464