I prefer putting the interfaces into the main library because there is one 
less step in between the composer and the music that way. Once the composer 
installs Csound 5, he or she is already able to call the Csound API directly 
from the composing language without any additional installation or 
configuration.

I have reconsidered the C++ interface. I do not think this is necessary. I 
think it is sufficient to provide wrappers for the existing C API.

I would like to see Csound callable directly from C (as it is now), Lisp, 
Python, and Java. Through Java, it becomes callable directly from 
Mathematica.

Regards,
Mike

----- Original Message ----- 
From: "Istvan Varga" 
To: 
Sent: Monday, October 10, 2005 3:34 PM
Subject: Re: [Csnd] Common Music and Csound 5


> Michael Gogins wrote:
>
>> As part of this work, I will be moving the C++ form of the Csound API
> > from the CsoundVST project into the Csound API library proper, and also
> > moving the Python wrappers.
>>
>> This will NOT introduce any additional dependencies in the Csound API
> > library build except for SWIG, and I can book in the generated wrapper
> > code so that even SWIG is not required for the normal build.
>
> Why is it absolutely necessary to integrate the C++ and other wrappers
> into the main library ?
> -- 
> Send bugs reports to this list.
> To unsubscribe, send email to csound-unsubscribe@lists.bath.ac.uk
> 

Michael Gogins wrote:

> I prefer putting the interfaces into the main library because there is 
> one less step in between the composer and the music that way. Once the 
> composer installs Csound 5, he or she is already able to call the Csound 
> API directly from the composing language without any additional 
> installation or configuration.
> 
> I have reconsidered the C++ interface. I do not think this is necessary. 
> I think it is sufficient to provide wrappers for the existing C API.
> 
> I would like to see Csound callable directly from C (as it is now), 
> Lisp, Python, and Java. Through Java, it becomes callable directly from 
> Mathematica.

This still does not explain why it should be made impossible to just build
the library with the C interface without including any wrappers, not depending
on more than an ANSI C compiler and libsndfile. Currently there is no use of
C++ in the main library, either.