Re: autoconfiscation -- questions about make usage

[Nathanael Nerode <neroden at doctormoo dot dyndns dot org>]

On Wed, 30 Jan 2002, mike stump wrote:
> > Hmm... where the heck do I find a free description of what a make has to
> > do to be POSIX compliant?
> 
> If you don't know this stuff stone cold...  It will make working on
> this harder.
> 
> Kinda like writing portable code, but not knowing what portable code
> is.  It is relatively hard for us to just say.
> 
> One pragmatic way, is to only use constructs that you see in the
> existing makefiles.

Yup.  Well, the existing makefiles are not SUS compliant, but are pretty
close.  (They use dashes and .PHONY.)  So I'll try to stick to SUS +
dashes + .PHONY.  

This means I am probably going to be writing a small program to
automatically generate Makefile.in, because there will be huge
lists of almost identical targets (which could be collapsed with pattern
rules, but not suffix rules).  This is an issue with any Makefile which
recursively makes subprojects, so I'm thinking of making a fairly generic
script.  (Unfortunately autoconf2.13 + automake is not sufficient, but
what I'm writing could probably be folded into one or the other at
some later time.)

If anyone's interested in seeing the preliminary version of RAM
('recursive automake'), tell me and I'll put it up somewhere once I've
actually written it.  (I wrote a spec so far, but not the actual
program... I haven't decided quite what form to make it in, and I'm only 
98% sure it's worth the bother.)

--Nathanael Nerode