On Tue, 20 Mar 2018, Russell Shaw wrote:

If autoconf'd programs required the end users to have an installation-shell that is ported to all systems of interest, then the learning requirements of autoconf users should be a lot less.

This has all been discussed many times before. If it had been decided that installing some other package would be a prerequisite for running configure, then this issue would not exist. However, that decision was never made. The precursor to Autoconf was Perl's Configure, which was/is shell-based.

If some other package could be a prerequisite then making the prerequisite a Bourne like shell with a m4 pre-processor would be a silly choice since shell scripts are a very poor choice to meet configure's requirements.

A specially-crafted language syntax would be a better choice to meet configure requirements in order to avoid obscure and inefficient m4/shell/sed syntax.

Regardless, Autoconf ultimately works well and the shell script part of the effort that a configure script implementor must create is usually not very difficult at all.

Bob
--
Bob Friesenhahn
bfrie...@simple.dallas.tx.us, http://www.simplesystems.org/users/bfriesen/
GraphicsMagick Maintainer,    http://www.GraphicsMagick.org/

_______________________________________________
Autoconf mailing list
Autoconf@gnu.org
https://lists.gnu.org/mailman/listinfo/autoconf

Reply via email to