🚀 go-pugleaf

> How committed is the GNU project to using the ./configure command
> prior to make?

Get standards.text from prep.ai.mit.edu.

> I have developed an alternative porting scheme that does not require a
> separate step before running make.  I simply require an environment
> variable DIALECT to be set to a name describing your environment.
> An include file details the requirements for the various environments.

Well, there's no harm in describing it, whether or not anyone decides
to use it eventually.  We're not against real improvements.

How would DIALECT be set?  What if you're on a homogenous network,
where you need to compile software for multiple platforms?
Cross-compiling?  Then I don't see that

DIALECT=sun3-os4-gas-x make

is easier or simpler than

configure +gas +x sun3-os4; make

bash and emacs currently do similar things, by doing ugly things with
Makefiles and the C preprocessor.

What happens when a new version of Ultrix (like 4.0), or OSF/1 comes
out, etc.?  You have to revise DIALECT and add new cases to your
include file.  Whereas if you just test for features, it's more likely
to work.

And how to you handle compiling in a different directory from the
source (the configure +src option) so you can have binaries for
multiple architectures around (and on local disks, with the source NFS
mounted) at once?
--
David J. MacKenzie <djm@eng.umd.edu> <djm@ai.mit.edu>