So I ported the C API for MathGL to D, and it is up at code.dlang.org (under dmathgl).  MathGL is a nice plotting library.

http://mathgl.sourceforge.net/doc_en/Pictures.html#Pictures

Later I will work on porting the C++ interface, but so far it at least works for the simplest sample.  (Not tried anything else as they are all written in C++).

DMD gave me an error message for the following declarations:

double mgl_rnd (...);
double mgl_rnd_ (...);

It says I need at least one fixed argument first.  But the C headers are as they are.

I could work around this by writing a C stub

double mgl_rnd_(int dummy, ...)

but is there any way to call these functions without this workaround?

sorry - posted in wrong forum

On Wednesday, 18 March 2015 at 00:54:07 UTC, Laeeth Isharc wrote:
> So I ported the C API for MathGL to D, and it is up at code.dlang.org (under dmathgl).  MathGL is a nice plotting library.
>
> http://mathgl.sourceforge.net/doc_en/Pictures.html#Pictures
>
> Later I will work on porting the C++ interface, but so far it at least works for the simplest sample.  (Not tried anything else as they are all written in C++).
>
> DMD gave me an error message for the following declarations:
>
> double mgl_rnd (...);
> double mgl_rnd_ (...);
>
> It says I need at least one fixed argument first.  But the C headers are as they are.
>
> I could work around this by writing a C stub
>
> double mgl_rnd_(int dummy, ...)
>
> but is there any way to call these functions without this workaround?

"Laeeth Isharc"  wrote in message news:jmlgralvzaqperfkntqw@forum.dlang.org...

> DMD gave me an error message for the following declarations:
>
> double mgl_rnd (...);
> double mgl_rnd_ (...);
>
> It says I need at least one fixed argument first.  But the C headers are as they are.
>
> I could work around this by writing a C stub
>
> double mgl_rnd_(int dummy, ...)
>
> but is there any way to call these functions without this workaround?

You should just declare them with the actual argument types.  From a quick look at the documentation it seems like the correct signature would be
double mgl_rnd();

i.e. these functions aren't actually variadic they just haven't specified the signature.  Functions that take variadic arguments will pretty much always have at least one named argument. 

On Wednesday, 18 March 2015 at 03:14:30 UTC, Daniel Murphy wrote:
> "Laeeth Isharc"  wrote in message news:jmlgralvzaqperfkntqw@forum.dlang.org...
>
>> DMD gave me an error message for the following declarations:
>>
>> double mgl_rnd (...);
>> double mgl_rnd_ (...);
>>
>> It says I need at least one fixed argument first.  But the C headers are as they are.
>>
>> I could work around this by writing a C stub
>>
>> double mgl_rnd_(int dummy, ...)
>>
>> but is there any way to call these functions without this workaround?
>
> You should just declare them with the actual argument types.  From a quick look at the documentation it seems like the correct signature would be
> double mgl_rnd();
>
> i.e. these functions aren't actually variadic they just haven't specified the signature.  Functions that take variadic arguments will pretty much always have at least one named argument.

Thanks.  I should have double checked but trusted dstep which
seems to have gotten confused by these ones.  Its a great time
saver generally though.


Laeeth

On 2015-03-18 05:49, Laeeth Isharc wrote:

> Thanks.  I should have double checked but trusted dstep which
> seems to have gotten confused by these ones.  Its a great time
> saver generally though.

Please report any issues with DStep to [1].

[1] http://github.com/jacob-carlborg/dstep/issues

-- 
/Jacob Carlborg

On Wednesday, 18 March 2015 at 16:14:28 UTC, Jacob Carlborg wrote:
> On 2015-03-18 05:49, Laeeth Isharc wrote:
>
>> Thanks.  I should have double checked but trusted dstep which
>> seems to have gotten confused by these ones.  Its a great time
>> saver generally though.
>
> Please report any issues with DStep to [1].
>
> [1] http://github.com/jacob-carlborg/dstep/issues

Was just about to ;)