add feature detection macro MP_HAS (2)

[minad]

See #214. I wanted to try again if I can make this work without using variadic macros, this time only relying on C89.

@karel-m Could you please try again your compiler suite?

@sjaeckel @czurnieden This PR is simpler and better than the one before. Hopefully it also works on older compilers. This helps to clean up control flow. Another advantage is that the compiler will check all branches. Dead code elimination will make sure that the dead branches are not compiled.

[minad]

@sjaeckel What do we have to check here to make sure that this change does not break any things? I don't think it does, but what do you guys need to be convinced 😉

If you are worried about dead code elim, try this with gcc -o dead dead.c

/* dead.c */
#include <stdio.h>

void does_not_exist(void);

int main(void) {
        if (sizeof("") == 1) {
                puts("We're ok");
        }
        if (sizeof("X") == 1) {
                does_not_exist();
        }
        return 0;
}

[minad]

In contrast to the solution in #214, this works on MSVC 2015 and 2017 as I just tested using appveyor.

[sjaeckel]

@karel-m can you please have a look if this works for your cases as well?

[minad]

Which is exactly what I would call the slightly growing lack of attention to unusual compilers.

This is what I call "a typo", nothing to complain about. Thank you for the fix.

[karel-m]

This is what I call "a typo", nothing to complain about.

Obviously. What I wanted to say is that nobody tried any compiler that hits #else branch.

[minad]

@sjaeckel Ping. See also https://www.gnu.org/prep/standards/standards.html#Conditional-Compilation where c conditionals are recommended for conditional compilation.

[minad]

@sjaeckel What is your opinion about this PR now?

[sjaeckel]

Obviously. What I wanted to say is that nobody tried any compiler that hits #else branch.

I'm not sure where the culprit has to be searched there... in the compiler providers who mostly try to mimic gcc or the ones who don't mimic gcc not being available easily...

Which compilers did we test this now with?

• gcc down to 3.4
• clang multiple versions
• MSVC multiple versions
• IBM xlc 11.1

Something missing? If this works with all those compilers I think it's safe to merge.

[minad]

@sjaeckel yes, the compilers you listed are the ones we tested so far.

[minad]

@sjaeckel I rebased this PR again on develop. CI all green.

[minad]

Ping @sjaeckel. Could you merge this PR or are there things to do?

[sjaeckel]

I still don't really like those complex if-elseif statements/chain e.g. in mp_mul() but I could live with it...

@czurnieden anything to add?

[czurnieden]

@czurnieden anything to add?

To mp_mul?
Yes, but not today, that'll need a bit longer ;-)

Otherwise: no problems, can tick it off, too.

[minad]

@czurnieden Thanks!

@sjaeckel

I still don't really like those complex if-elseif statements/chain e.g. in mp_mul() but I could live with it...

Yes, it might be possible to improve this. In order to not introduce any unwanted changes, I took the most direct translation from the intermingled #if and if branches.

[minad]

@sjaeckel ping?