gEDA-dev: Re: [Gnucap-devel] Gnucap docs build failure on FC5
(and other places)
al davis
ad106 at freeelectron.net
Mon Aug 7 20:43:14 EDT 2006
On Monday 07 August 2006 18:36, Dan McMahill wrote:
> Stephen Williams wrote:
> > al davis wrote:
> >>There is a real need for a high quality replacement for
> >>autoconf.
> >
> > I've had the name "CMake" wispered in my ear once or twice,
> > and I've been considering looking into it. Apparently, some
> > groups find it to be superior to autoconf, but I have no
> > idea what I think of it, since I haven't look at it yet.
>
> ack. I'll believe its better when I see it actually working
> on lots of platforms. Not saying it isn't but it surely is
> not as widely deployed as autoconf/automake.
It depends on your definition of "better". I looked at it
briefly, just enough that I don't want to look any more.
I still intensely dislike autoconf. CMake doesn't solve ANY of
the problems I have with autoconf.
I am tempted to take another look at the old gnucap build
system, which I kluged together over 20 years ago, and turning
it into a real build system. I think by doing this, stealing
the good parts from autoconf, it can be a system far better
than anything available.
Most people are not building super big projects, and don't need
the overhead. What is needed is something simple.
The old gnucap build system is written entirely in "make", and
should remain so. I suppose it is ok to use /bin/sh too.
(Notice I didn't say "bash".) It doesn't require any
non-source files to be distributed. It doesn't make a mess (by
default) of the source directory. It allows a single command
to make multiple builds without interfering with each other.
The only problem is that it doesn't have all the script options
people seem to need. As to portability, it still works, with
almost no changes, despite the fact that the floppy based CP/M
machine I used for the initial work is seriously obsolete. The
biggest change was the addition of support for hierarchical
directories, when MS-DOS-2 came out.
More information about the geda-dev
mailing list