[Numpy-discussion] Code for compatibility with gfortran, license issues

David Cournapeau cournape at gmail.com
Mon Jun 29 13:47:42 EDT 2009

On Tue, Jun 30, 2009 at 2:15 AM, Robert Kern<robert.kern at gmail.com> wrote:
> On Mon, Jun 29, 2009 at 00:34, David
> Cournapeau<david at ar.media.kyoto-u.ac.jp> wrote:
>> Hi,
>>    I started working on a new approach for windows 64 bits support, to
>> be able to combine gfortran and visual studio. Basically, I am
>> reimplementing the needed functions from libgfortran so that it can be
>> built with MS compiler, but I cannot hope to do that without using
>> gfortran sources (under the GPL 3), if only for the signature/API of the
>> functions.
> Function signatures and public APIs are not copyrightable. What else
> do you need?

I need to implement some intrinsics (_gfortran_pow_i4_i4, etc...). I
would not consider them public a priori, but I don't really know the
legal meaning of public API.

> Keep in mind that libgfortran is GPLv3+exception. Is the code that you
> are looking at from gfortran itself (GPLv3) or libgfortran

It should be only libgfortran (and libgcc) - I could not quite
understand the exception part for libgfortran and its implications.

One function which may be problematic is  __chkstk. It seems that
anything different from the exact implementation would be troublesome.
It is implemented purely in ASM, and does not use standard calling
convention, so anything else than a verbatim copy sounds dangerous
(the function itself is like 5-6 instructions). I could get rid of
most non-computational functions by modifying the fortran code in
scipy, except for this one.


More information about the NumPy-Discussion mailing list