[Ecls-list] ECL + ContextL?

[Alexander Gavrilov]

On Tue, Nov 3, 2009 at 3:46 PM, Waldek Hebisch <hebisch at math.uni.wroc.pl> wrote:
> Closure CL allows functions written in Lispy assembler.  I am
> not sure if you consider this reasonable (I did not check,
> but I think you can generate Lispy assembler from macros).

I'd consider it reasonable if it would manage stuff like register
allocation and spilling to stack for me, like the C SSE intrinsic
interface does. Writing my own register allocation engine is really
too much work.

Generally the most difficult part in full SSE intrinsic support is
that it requires adding a new special 128-bit atomic immediate type to
the compiler, with full support for type inference, boxing/unboxing,
allocating temporaries on the stack and in registers, etc. I wonder
how difficult might be doing that in ECL on top of declarations in
xmmintrin.h.

Alexander