[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Proposed comments and changes to P754 draft (by Bob Davis)



I think storage formats are basically a lost cause, because with very
narrow formats, the boundary between exponent and significand becomes
very important and application-dependent.     We have basic formats for
which we can argue that the boundary is properly fixed for the bulk of
applications.      Going wider or narrower, one gets into the realm
of application-dependent or platform-dependent.

So I'd standardize on basic formats: a conforming implementation provides
all the required operations on one or more basic formats, which are 
interchange formats in internal form.

Implementations may always provide additional formats; if they
provide the same operations as the basic formats, call them extended formats,
which are not interchange formats in internal form.

Converted to character sequences, all formats can be interchanged.

So standard formats have just two subsets, basic and extended.
Sound familiar?


754 | revision | FAQ | references | list archive