Die GNU autotools

[Tzafrir Cohen]

On Mon, Jan 10, 2011 at 09:02:15PM +0200, Elazar Leibovich wrote:
> You gave a very good history lesson about the need and reasons for
> autotools.
> 
> The thing is, many of the software written now, is not intended to run on
> HP-UX framework. I'm not a Unix expert, but maybe modern unices are more
> POSIX compliant than in the past.
> 
> So I think developers prefer to write a more strictly POSIX compliant
> application, and to need a simpler building process and have a single set of
> source files, than to use non-portable functions and to be forced to use
> autotools.
> 
> For example, I'd rather implement hton* myself, then checking for this
> function's existence with autotools.
> 
> A good example for such a project is http://re2.googlecode.com It uses plain
> makefile, supports a few unices, and is fairly complex.

Nice. 'make distclean; make' fails there. I had to see in the makefile I
need to enable something that requires internet access. Now that's
distro-friendly.

Looking at the makefile I noticed you use there 'ifeq ($(shell uname),Darwin)'

What if I want to cross-build? Do I have to override 'uname' in the
PATH?

In fact, the C++ compiler is g++ explicitly in some parts. Though you
define on the top CC=g++ (HUH? CC? Not CXX or CCC?) . But then again,
you use it to compile .c files as well.


Yeah. You have your own hand-crafted Makefiles. With the only
special-case of Darwin. No need to be portable.

-- 
Tzafrir Cohen         | tzafrir at jabber.org | VIM is
http://tzafrir.org.il |                    | a Mutt's
tzafrir at cohens.org.il |                    |  best
tzafrir at debian.org    |                    | friend