[MLton-devel] FFI real as float? Also a note to Anoq.

[Matthew Fluet]

> > I just checked in the updated x86-codegen and most of the work for a full
> > Real32.  The commit should be emailed, but I need to run now.  I just
> > wanted to add that the mods for Real32 are untested on Sparc.
>
> With make from clean sources, I get the following regression error.
>
> testing array
> x86Translate.translateChunk::x86Translate.Chunk.toX86Chunk::x86Translate.Block.toX86Blocks::x86Translate.Transfer.toX86Blocks::FIXME

It seems to be due to the fact that Primitive.Int64.toInt is declared as
_import "Int64_toInt32": int -> Int.int.  Somewhere in the basis (I
think in integer.fun) are a couple of conversions.  For some reason, they
stick around through dead code elimination; unfortunately, this means that
as _import's, they can't be dropped.  Thus, they appear in the final SSA
program (albeit, with no destination).  The x86-codegen chokes when it
tries to ask for the size of the destination, based on the functions
Creturn type.

I think the simplest solution is to turn the _import's into _prim's;
again, the x86-codegen will choke if those prim's make it all the way to
the codegen, but tht shouldn't happen.



-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01
_______________________________________________
MLton-devel mailing list
MLton-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mlton-devel