Re: [Gcl-devel] RE: [Maxima] Re: Maxima cmucl patch

[James Amundson]

On Fri, 2004-01-30 at 11:49, Camm Maguire wrote:

> As for developers, there are competing concerns -- designing a
> self-contained build process could attract people unused to the GNU
> toolchain, but having to reinvent all that functionality ourselves and
> for each project (i.e. gcl, maxima....), or to maintain a fork kept in
> sync with a traditional unix-like autoconf setup, would seem rather
> onerous given our current resources. 

I'll throw in my two cents here. Since Maxima is written in Lisp, I have
relied on the standard Lisp tools where appropriate. My goal for Maxima
was to make building and installing Maxima comfortable for someone who
is familiar with building and installing software under Unix, as well as
making the Lisp part comfortable for experienced Lisp users.  My latest
efforts (which lead to this thread) are only to make good on the latter
part.

I can make the Maxima build a purely lisp-based operation with minimal
duplication of autotools/automake code because Lisp is a very
self-contained environment. There wouldn't be an analogous approach for
the GCL build.

> Is using/installing MSYS really difficult for a 'windows
> developer'?

Empirically, I surmise that many people who like Windows HATE using
MSYS. It make sense to me -- people who like that sort of environment
usually run Linux (or other Unix-like OS).

> In like manner, I've never understood the efforts by some to reproduce
> the functionality of 'make' within lisp itself (i.e. defsystem??)  In
> other words, once we have a working tool, why not just use it, even if
> 'not invented here', and go on to the business at hand?  Provided of
> course that said tool is not too difficult to obtain?

Even though I hate to see people re-re-reinventing the wheel. I don't
think that the case with make vs. defsystem. They are not comparable
tools.

--Jim