Re: Ada.Strings.Fixed.Count raises Storage_Error

["Jeffrey R. Carter" <spam.jrcarter.not@spam.not.acm.org>]

On 06/29/2016 11:49 AM, Niklas Holsti wrote:
> 
> I'm astonished that you seem to be saying that Gödel's results mean that this
> clear and simple bug in a library function should not be reported to AdaCore.
> Really. Perhaps you are afraid that this is the last bug in GNAT, and as you
> believe that every compiler must have at least one bug, removing this bug would
> mean that GNAT would no longer be a compiler :-)

This is not an error in the compiler; it's an error in a general-purpose
library. Fixing it will not change the number of errors in the compiler.

We're told that the ARG has decided that some corner cases are too difficult for
compiler writers to get right, so the ARG won't require that they be handled
correctly. The argument that this permission means the function in question
doesn't need to be correct is wrong because the error in question isn't
something the compiler has difficulty compiling. It's an error by the
implementor of the subprogram that the compiler compiles correctly.

The argument that correcting the error would have a negative effect on the
performance of the subprogram, and that it's more important that it be fast than
that it be correct is wrong because correctness is always more important than
performance for a general-purpose library. It's expected that a general-purpose
library's performance may not be acceptable for some applications. Such
applications know that a general-purpose library may not meet their performance
needs.

-- 
Jeff Carter
"What's the amount of the insult?"
Never Give a Sucker an Even Break
104