Re: Integers and Mathematical Correctness

["Eric G. Miller" <egm2@jps-nospam.net>]

In <3BE766F6.3D014CD8@worldnet.att.net>, James Rogers wrote:

> "Eric G. Miller" wrote:
>> 
>> #include <float.h>
>> #include <limits.h>
>> #include <math.h>
>> 
>> In C: sizeof float <= sizeof double <= sizeof long double
>> 
>> It's all there.  Granted, there are a few things where there's no analog
>> in C that you'd have to code/check yourself (limited ranges, fixed
>> precision).
> 
> The include files listed above declare constants for various values, such
> as FLT_MAX. In one sense they are like 'First and 'Last in that they
> identify value boundaries. Unfortunately, these values in C are not as
> useful as their Ada counterparts.
> 
> Take the following C snippet:
> 
> float f = FLT_MAX;
> double d = 2;
> 
> f *= d;
> 
> What will the result of this be? There is no error issued either at
> compile or run time. The value output by printf using a gcc compiler is
> "1.#INF".

Yes, you have to do sanity checking at run time.  C promotes all sorts
of unsafe things (as you well know).  Funny, I ran the snippet through
lint (default settings), and it didn't see an error either (nor does
-Wall generate any warnings).

[snip]

> Note that this error would have been caught by an Ada compiler. C never
> notices the problem at all.

This is one of the things that drew me Ada.  It's taking some getting used
to, but I think it's worth it.  I was shocked (and somewhat annoyed) by
how much the compiler complained when working on my first program.  The
programs are better because of it, though.