On Tue, 2026-03-03 at 08:14 +0100, Janneke Nieuwenhuizen wrote:
> We've had long-standing mingw branches that mostly Mike Gran and I would
> work on, now and then, e.g.,
> 
>     https://codeberg.org/janneke/guile/commits/branch/wip-mingw-guile-2.2
> 
> Almost excatly five years ago, I created a 64-bit port
> 
>     https://codeberg.org/janneke/guile/commits/branch/wip-mingw-3.0.8
> 
> on top of Mike's recent work with help from him and Andy, but it had an
> awfully large patch 
> 
>    
> https://codeberg.org/janneke/guile/commit/91d27b579f58b8aa214912fd7324747b248b0ad4
> 
> and never got merged.  So we've just been patching Guile ourselves (to
> create Windows binaries for Dezyne, for example).  Recently, Jonas
> Hahnfeld managed to split that patch up in several bits, make fixes for
> lightening, and get it merged; lovely!

I would like to clarify that I *did not* split up the patch because it
was entirely based on the concept of requiring and patching mini-gmp.
The merged changes work with upstream interfaces of GMP, at the expense
of a slightly slower conversion in case the value at run-time falls
between 2**32 and 2**64 and does not fit into long.

> We're still carrying a number of patches, e.g., on
> 
>     https://codeberg.org/spk121/guile/commits/branch/wip-mingw-2026
> 
> and
> 
>     https://codeberg.org/janneke/guile/commits/branch/wip-mingw-2026
> 
> some of which (compile time file-name-convention for MinGW) are too
> broken and ugly to merge and for which there's no solution yet.

Some of the changes may still be useful, but I would like to ask that
each of them is checked, to avoid that we inadvertently introduce
downstream modifications to mini-gmp (which should not be needed
anymore).

Jonas

Attachment: signature.asc
Description: This is a digitally signed message part

Reply via email to