RE: [Axiom-developer] RE: algebra Makefiles with explicitdependencies, bootstrap, fixed-points etc.

[Bill Page]

Steve.

On Sunday, January 09, 2005 10:54 AM you wrote:
> ...
> perhaps we should dump all the bootstrap code and update all
> pamphlet files?

Yes I agree. But see also below.

> 
> If you agree this is a good move, perhaps we can think of
> the possible advantage of automating the process. e.g. a make
> option to update the related bootstrap pamphlet files with
> fresh code. I have a feeling, however, that this is such a
> central aspect of the build it might better be left as a
> `by hand process'. What do you think?

I don't think there would be much point in automating this
separately from the issue of implementing a general "fixed
point" build process. In the current situation in Axiom it
seems that it is possible simply to update the existing set
of bootstrap files with the fixed point lisp code and thereby
guarrantee a fixed point solution in just one pass. But I
do not think that this is possible in general. I believe that
changes in some Axiom modules will eventually necessitate
a different selection of bootstrap files in order to obtain
a one pass build.

I will write a little more about this in reply to your other
message.

> 
> I am willing to do this work today if you like. Tim, would
> you prefer to take this on given your expertise? If not, any
> words of wisdom? ;)
> 

I don't think that replacing the bootstrap lisp code is difficult
or particularly subtle. All you need to do is copy the appropriate
*.NRLIB/code.lsp files back to the <<*.lsp BOOTSTRAP>> chunks in
the src/algebra/*.spad.pamphlet files. I think you should go ahead
and do this and update the axiom--main--1 branch. I can grab your
changes from there and apply them to the axiom--windows--1 branch.

But lets keep in mind that this is not a general solution. It
only solves the build problem and retains the current make process
for now. I think we should continue to pursue the idea of mutual
recursion in Axiom since it is (at least implicitly) such an
essential part of the current Axiom library.

Regards,
Bill Page.