Re: latest gcc vs lib/timespec.h:85

[Tim Rühsen]

On 12/13/2017 10:55 PM, Paul Eggert wrote:
> On 12/13/2017 01:32 AM, Tim Rühsen wrote:
>> Now clang throws out an annoying warning about the return value of  >
>> timespec_cmp(): > > In file included from wget.c:51: > 
> ../lib/timespec.h:94:20: warning: implicit conversion loses integer >
> precision: 'long' to 'int' [-Wshorten-64-to-32] > return a.tv_nsec -
> b.tv_nsec; > ~~~~~~ ~~~~~~~~~~^~~~~~~~~~~ > > I wonder if we can't
> silence clang and gcc by keeping the 'assume()' > *and* using return
> (int) (a.tv_nsec - b.tv_nsec));
> I'd rather continue to omit the cast, as casts are too powerful (it's
> too easy to get them wrong, with no diagnostic).
> 
> -Wshorten-64-to-32 is like -Wconversion, and we should ask people not to
> use -Wshorten-64-to-32 in the same way that we ask them not to use
> -Wconversion. Does it fix things to add -Wshorten-64-to-32 to
> build-aux/gcc-warning.spec and to build-aux/g++-warning.spec?

No, it doesn't change anything (I am not using manywarnings.m4).

What about a #pragma here ?

And of course I can disable the warning for the gnulib code alone...
that isn't the point.
Switching on warnings for the gnulib code was meant as a help in the
means "more eyes see more things". It was once was a developers who
asked for not switching off warnings in gnulib code and I agreed. But I
don't definitely don't want to waste both your and my time with nitpicking.

With Best Regards, Tim

signature.asc
Description: OpenPGP digital signature