I believe I should roll back some of the changes in the j+gmp-clean
branch of jsource.

The issue shows up in test/g212.ijs where some tests fail.

A deeper cause is that to expedite windows development and portability
work, we tried using a windows specific fork of libgmp named mpir,
because it was "more windows friendly" than libgmp. And, as a
consequence, changes were made to vx.c which prevented compilation
under linux.

Specifically, type I (long long) which is 64 bits wide was replaced
with type long which on 64 bit windows is 32 bits wide, for backwards
compatibility reasons.

That said, I think this also highlights that I should have gotten
libgmp working for both a 32 bit architecture and a 64 bit
architecture before asking other developers to look at the code.

I don't think this porting effort was entirely wasted. But I now need
to take at least some of the lessons learned here and apply them.

I'm going to have to do a significant rewrite of this code base. And,
when I open it back up as "alpha quality" it will be on a new branch.

FYI,

-- 
Raul
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to