Does anyone know:
If an extern(c) function is defined as returning a bool, does D handle the conversion automatically into true and false?  Should it be defined as returning  an int?  On a 64 bit machine?  Does this depend on the compiler used to compile the library?  int32_t? int_64_t?

Is this portable or non-portable?

-- 
Charles Hixson

Charles Hixson:

> Does anyone know:
> If an extern(c) function is defined as returning a bool, does D handle the conversion automatically into true and false?  Should it be defined as returning  an int?  On a 64 bit machine?  Does this depend on the compiler used to compile the library?  int32_t? int_64_t?
>
> Is this portable or non-portable?

A D bool is a C99 uint8_t that has values just 0 or 1.

If your uint8_t contains a value x > 1, in many cases this works, but some D code relying on the standard values of a D boolean breaks (like when you sum bool values in D, to count the true ones).

D doesn't handle those conversions beside the narrowing or extension of bit-width lengths.

Bye,
bearophile

On 2013-09-26 20:56, Charles Hixson wrote:
> Does anyone know:
> If an extern(c) function is defined as returning a bool, does D handle
> the conversion automatically into true and false?  Should it be defined
> as returning  an int?  On a 64 bit machine?  Does this depend on the
> compiler used to compile the library?  int32_t? int_64_t?
>
> Is this portable or non-portable?

In general 0 is false and all other integers are true. So if the function returns 0 or 1 it will automatically work for true and false as well.

-- 
/Jacob Carlborg

On 09/26/2013 12:19 PM, bearophile wrote:
> Charles Hixson:
>
>> Does anyone know:
>> If an extern(c) function is defined as returning a bool, does D handle the conversion automatically into true and false?  Should it be defined as returning  an int?  On a 64 bit machine?  Does this depend on the compiler used to compile the library? int32_t? int_64_t?
>>
>> Is this portable or non-portable?
>
> A D bool is a C99 uint8_t that has values just 0 or 1.
>
> If your uint8_t contains a value x > 1, in many cases this works, but some D code relying on the standard values of a D boolean breaks (like when you sum bool values in D, to count the true ones).
>
> D doesn't handle those conversions beside the narrowing or extension of bit-width lengths.
>
> Bye,
> bearophile
>
It's a library function, that claims to return bool, but is a C function.  My suspicion is that the proper way to handle this is to say that it returns an int, or possibly an int32_t, i.e. to declare is as not returning a bool, even though the library says that that's what it returns.  But my computer is a 64 bit machine, and the code was compiled awhile ago, so I suspect it may really be returning an int32_t.

The problem is that if I guess wrong this will not dependably cause a problem.  (I had a hope that D would say "O, it's a C routine returning a bool, so I know how to deal with that", but that wasn't really my expectation.)

Yeah, if I were defining the function, I'd define it in a way that I know how to handle properly.  But I'm not, and the interface is for either C or C++.   I can choose between those two.

So... if C returns an int64_t and I declare it as an int32_t, will this cause problems?  What about if C returns an int32_t and I declare it as an int64_t?  Or *is* there a good way to handle this? (I think that there's an error code I could interrogate if I need to just ignore the result...but I don't know if I can do this in all cases.)

-- 
Charles Hixson

Charles Hixson:

> So... if C returns an int64_t and I declare it as an int32_t, will this cause problems?  What about if C returns an int32_t and I declare it as an int64_t?  Or *is* there a good way to handle this? (I think that there's an error code I could interrogate if I need to just ignore the result...but I don't know if I can do this in all cases.)

Take a look at this entertaining page:
http://dlang.org/phobos/std_stdint.html

Bye,
bearophile

I'd forgotten about:

	int_least8_t

I guess that's what I should use.

On 09/26/2013 03:06 PM, bearophile wrote:
> Charles Hixson:
>
>> So... if C returns an int64_t and I declare it as an int32_t, will this cause problems?  What about if C returns an int32_t and I declare it as an int64_t?  Or *is* there a good way to handle this? (I think that there's an error code I could interrogate if I need to just ignore the result...but I don't know if I can do this in all cases.)
>
> Take a look at this entertaining page: http://dlang.org/phobos/std_stdint.html
>
> Bye,
> bearophile
>


-- 
Charles Hixson