RE: [Gcl-devel] 2.6.2 - proposed patch for pcl/makefile

[Mike Thomas]

Hi Camm.

| > ========================================================================
| > RCS file: /cvsroot/gcl/gcl/pcl/makefile,v
| > retrieving revision 1.5.6.2
| > diff -r1.5.6.2 makefile
| > 22c22
| > <       '(setq compiler::*keep-gaz* t)'
| > ---
| > >       '\#+winnt (si::use-fast-links nil) (setq
| compiler::*keep-gaz* t)'
| > ========================================================================
| >
|
| This is fine too.  I'm wondering how we are going to remember to
| remove this once its fixed.

I wondered that myself, but then remembered that this bug has sucked out
most of my spare programming time for over a year now to the detriment of a
number of projects which I want to work on.

I can assure you that I will not forget if this bug is fixed.


| We need to think of a way to systematize
| parameters like this in a single minimal place.

Unless you happen to mean the location in which the CL symbols such as
winnt, gcl, kcl etc are set up in the source, then your suggestion suggests
configure.in to me - I wanted to avoid making changes there so close to a
release and when there is still the possibility that we might manage to
squash the bug.

While thinking about configure.in in this context, I also wondered whether
we shouldn't have defsystem as a permanent fixture in GCL and use it in our
build.  It seems to be a feature of some major Lisp implementations and one
of the two main frustrations I have had with using GCL on third party code
has been its absence - the other being the absence of CLOS and ANSI.

In particular, I would feel that we had reached a very useful level of
conformity if we could load, compile and run the CL-PDF library and friends
without making changes to the code:

   http://www.fractalconcept.com/

Cheers

Mike Thomas.