[Axiom-developer] axiom--windows--1--patch-10

[Bill Page]

Tim,

I have commited recently two new patches for the windows
branch.

patch-9 applies 5 patches from Savannah. There are still
a few outstanding proposed patches but I still think these
others need to be investigated more before commiting them.

patch-10 implements several new implicit make rules in the
src/Makefile.pamphlet, src/algebra/Makefile.pamphlet,
src/input/Makefile.pamphlet and src/etc/Makefile.pamphlet.
The main reason for these is because of the continuing
problems that I have had trying to get a clean build of
the Savannah patches to axiom--main--1. I just got very
very frustrated with the bloated structure of the Makefiles
(some with more than 40,000 lines of code!) and my ability
to do everything right. As per our previous discussion you
agreed to a compromise in the design of the Makefiles to at
least use implicit rules where possible. I think I have
solved the problem you mentioned in algebra/Makefile.pamphlet
concerning your first attempt to use implicit rules.

In axiom--windows--1--patch-10 I have implemented these new
implicit rules and this has reduced the Makefile.pamphlet to
not more than 1,600 lines including the documentation. The
reason that I applied this change first to axiom--windows--1
is that I knew you were about to make significant changes to
axiom--main--1 for hyperdoc and I thought it best not to mix
problems and also I have been having much more success with
tla on the axiom--window--1 branch.

For me, even though I know the files are highly structured
and you have gone to some effort to describe that strucuture
in the Makefile.pamphlet files, there is a BIG difference in
how easily I can appreciate and maintain the structure of a
file with 1,600 lines of code versus one that is over 40,000
lines of code.

Anyway, when you get a chance, I would like to get your
opinion of these changes to the Makefiles. Do you find them
clear and otherwise acceptible. If so, I will commit the
same changes to the axiom--main--1 branch along with my
Savannah patches.

Also, although I have not commited any changes for this yet,
I have also been working on adding the missing dependencies in
the algebra/Makefile. I have a very simple script that extracts
this dependency information from the ``Loading'' messages in
the output of the )compile commands. Although it will increase
the size of the Makefile again, adding this information to the
algebra/Makefile will allow changes to spad files to properly
propagate to other files in order to minimize the total number
of files that have to be re-compiled.

Regards,
Bill Page.