Re: source(builtin) and read(2)

[Geoff Clare]

Matthew Woehlke <address@hidden> wrote, on 30 Mar 2007:
>
> Geoff Clare wrote:
> >The code was just to illustrate the point that if it is possible for
> >the condition (var > SSIZE_MAX) to be true then the implementation
> >does not conform to the requirement that SSIZE_MAX is the maximum
> >value of an object of type ssize_t.
> 
> I still don't buy that. If that was the intended meaning then I would 
> expect the standard to read "...maximum value of a ssize_t" or "maximum 
> value of *type* ssize_t". It doesn't. It uses the word "object" in this 
> one, and ONLY one, instance. If this was *not* done to create a 
> distinction between representable and permissible values, then why *was* 
> it done?

The wording difference is accidental.  The 1990 standard did not
describe CHAR_MAX, INT_MAX, etc. at all, it just listed them as
limits from the C Standard that were required by POSIX.1.  Thus
SSIZE_MAX was the only limit of that kind that had a description
in POSIX.1-1990.

There are variations in the wording for the other limits as well.
E.g. CHAR_MAX is "Maximum value of type char", but INT_MAX is
"Maximum value of an int".  There is no significance to any of
these variations.

Interestingly the wording for SSIZE_MAX in 1990 was slightly
different: "The maximum value that can be stored in an object of
type ssize_t".  The current wording seems to have come from XPG4.
Since XPG4 deferred to POSIX.1-1990, there would have been no
intended difference in requirements.

-- 
Geoff Clare <address@hidden>
The Open Group, Thames Tower, Station Road, Reading, RG1 1LX, England