Re: Languages, languages ... (was: 754R and Reproducibility)

[Guillaume Melquiond <guillaume.melquiond@xxxxxxxxxxx>]

Le mercredi 11 juillet 2007 à 03:59 -0400, Michel Hack a écrit :

> It is true that a specification at the level I just described is not what
> most programmers want (basically, an assembly-language level), but I claim
> that it is necessary to enable a small group of programmers to provide the
> conformance at the higher levels (basically, compiler writers).  So I am
> not at all opposed to providing clear conformance requirements at the
> language level (which was missing in 754(1985)) -- what I'm complaining
> about is dropping conformance discussions at the lower levels altogether.

I may be misunderstanding. But it seems like you would like to require a
set of floating-point operations to be available directly at
assembly-language level. I think this is a really drastic move.

For example, there are some DSP processors that have integer arithmetic
units only, yet the platforms they are embedded into provide a software
implementation of floating-point arithmetic (thanks to some help from
the processor, like the ability to count leading zeros, performances are
not as bad as one may expect). Such a platform could never claim
conformance with the standard, as absolutely nothing would be available
at assembly-language level.

Another example is some modern generic processors, which do not provide
division and square root any longer. Although they are first-class
citizens of the standard, these operations have to be implemented in
software. It seems to me that lots of people agree that it was a smart
move to remove them from hardware. With your criteria, would such
platforms (processor+compiler) still be able to claim they provide
compliant floating-point arithmetic?

So, could you detail the features you expect from the lower levels?

Best regards,

Guillaume