🚀 go-pugleaf

>>>>> On 13 Feb 92 14:28:04 GMT, moss@brl.mil (Gary S. Moss <moss>) said:

Gary> I think that depending on environment variables is a bad idea.
Gary> The source code should compile correctly regardless; it is easy
Gary> enough to define the compiler and loader flags in the makefile,
Gary> cakefile, Imakefile, or whatever you use.  In fact, good
Gary> software distributions take care of this with a top-level
Gary> configuration file.  It is unlikely that using environment
Gary> variables provides flexibility, typically you always use the
Gary> same include and lib directories to compile a given program.
Gary> Also, it would be a pain to have to change environment variables
Gary> when compiling different executables, possibly in the same
Gary> source tree.

I think environment variable hooks should be provided whenever a
software system has to know about pathnames for finding include files,
data files, libraries, compiler passes, auxilliary programs, etc.
For example, the Mit X11R4 sample distribution compiled for installation
in "the usual places" (under /usr) can be made to work properly
if you put it elsewhere by means of environment variables.  If
you want to install things in non-standard places because of
disk space or access restrictions or NFS server politics, this is
very important.

Top level config files that allow for setting pathnames at compile
time are also essential, but not a substitute for environment variable
hooks.  What if you want to change the location of the installed files
because you have access to a new disk or a new server?  You don't want to
have to recompile all your software, so its nice if you can just set
some new environment variables