Re: [sv-bc] question about integer expression

John Havlicek wrote:
> Can anyone comment on whether "integer expression" is intended to
> limit the result of the expression evaluation to what can be stored in
> the 32-bit integer data type?
>
> I think that the answer should be "no".
>   
John,

    I also believe the answer is "no" for the purpose of establishing
what such constructs mean,
so for example there is no 32-bit "context" to possibly affect the
arithmetic of those self-determined
expressions.  Similarly I read identities like the A[E+:W] <=> A[E:E+W]
that Arturo cites as ideals
rather than literal formulas (E and W's types never interact).  One I
had to cope with recently is
the idea that declaring C[expr] means C[0:(expr)-1] - don't take that
one too literally either!

But the "no" comes with a qualification - an economical implementation
usually has limits when
trying to serve extremes here.  While there is no semantic bias toward
32-bit arithmetic, there are
huge practical biases.  A portable RTL design would be foolish to count
on any larger range.
Prescriptively, the standard should say "integral expression".  But, in
conforming RTL, any
index or bounds value should be representable as a native integer on the
host machine.

Greg

Disclaimer: JMHO -  not a voting position.



-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.