[Gcl-devel] GCL 2.6.6 does build Axiom on Windows

[Bill Page]

Mike, Camm,

On Tuesday, February 15, 2005 10:18 PM Mike Thomas wrote:
> 
> Good news I think.

Yes, I can confirm that it *does* build Axiom! It was my error.

> 
> I end up with a directory called "axiom--windows--1--patch-13"
> in which I successfully built Axiom on Windows:
> ... 
>   2. using 2.6.6 after modifications to "lsp/Makefile.pamphlet"
> for minimum usual patching to the GCL 2.6.6 source in new
> version gcl2.2.2w.
> 

This is exactly the same thing I did.

> The modifications I made apart from adding "gcl-2.6.6.tgz" 
> and new patch files in "zips" are to be found at the end of
> this email.

Comparing my tla file-diffs with yours, however I found that
I had unintentionally also removed the mingw.defs patch. :(
Thanks very much for helping me to sort this out.

> I believe I am using gcc 3.4.2, binutils 2.15.91, mingw32
> runtime 3.5, win32 libraries 3.1.

Yes, same.

> 
> Interestingly I get different output (less verbose) at the point
> where AHYP.spad is compiled to that you mentioned in your earlier 
> email, Bill, and to which I was previously accustomed:
> 

Yes. In this version of the Axiom makefile the output of the
algebra compiles is redirected to a file with the name int/*.out
so that the compiler output for AHYP.spad is in 'int/AHYP.out'.
I did this so that I could process the compiler "Loading"
messages to produce a list of dependencies. If anyone is
interested I can send you a copy of the script that I use to
extra these dependencies - but that is a whole other issue
since the list of dependencies is circular. Anyway I think
this approach is more useful than just adding the compiler
output to the other noise generated during the build.

You might also be interested in the 'fixedPoint' script which
is included in axiom--windows--1--patch-13. This script
resolves the inconsistencies between the algebra bootstrap
lisp code and the code produced by the Axiom compiler.

Regards,
Bill Page.