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

[Guillaume Melquiond <guillaume.melquiond@xxxxxxxxxxx>]

Le mercredi 11 juillet 2007 à 01:22 -0400, Michel Hack a écrit :
> > Now I am even more confused.  For example, take the soft-float library.
> > ...        Thanks to this library and your proposal, it means that any
> > vendor can claim to produce a processor compliant with the standard.
>
> This is true even without my proposal:  The combination of any hardware
> plus soft-float, assuming soft-float does support that hardware, would
> indeed be compliant.

I don't agree it is true without your proposal. The current draft talks
about compliance at the language level. In particular, if your language
does not give you (transparent) access to the features of the soft-float
library, then you cannot claim that your processor, even though
soft-float is available, is compliant. Practically, "gcc -msoft-float"
gives you a compliant environment in C, but it is possible that "gcc"
alone does not, so your processor vendor should not be able to claim
compliance.

In other words, with your proposal, any processor is compliant, even if
it only provides integer units. But with the current draft, only a
combination of compiler+system+processor that provides floating-point
arithmetic to the developer can be compliant. The vendor of a Turing
machine cannot claim to be compliant with the standard, even if it
provides a manual describing a state automaton that computes
floating-point operations.

In my opinion, that is the whole point of setting the standard at the
language level: what the standard describes is what the user is provided
with, and not what the processor may provide.

Best regards,

Guillaume