GMP 5.1.1 make check fail on Solaris 11

Marc Glisse marc.glisse at inria.fr
Fri Feb 15 10:44:59 CET 2013


On Fri, 15 Feb 2013, Niels Möller wrote:

> Torbjorn Granlund <tg at gmplib.org> writes:
>
>> Since this is common,
>> perhaps we should add an initial test t-does-exception-work-at-all.cc?
>
> Or a configure check?
>
> How essential are exceptions to gmpxx?

They are not really essential. If you don't try to create numbers from 
invalid strings, they won't happen. But some programs may depend on them: 
ask a user for a number, read the string, try to create a number from it, 
and if you catch an exception complain to the user and try again (instead 
of crashing). Some software relies on exceptions in an unsupported way to 
catch allocation failures.

> Would it make sense to completely
> disable building of the C++ stuff if exceptions are not working, or is
> it possible to just disable exception-related features and tests?

I think it is best not to disable anything. And if your compiler is 
broken, if an error occurs, the program will be terminated instead of 
letting you catch the exception. It seems ok to me. For make check, I 
think users would like to know that their compiler is that badly broken.

-- 
Marc Glisse


More information about the gmp-bugs mailing list