SWIG and Ruby/DL

[Peter C. Verhage]

Hi,

Just thinking out loud, but wouldn't it possible to generate Ruby/DL
code instead of Ruby C wrapper code using SWIG? This would make it a lot
more easy to use extensions cross-platform without the need for compilation.

Regards,

Peter

 

[Jamis Buck]

Hi,

Just thinking out loud, but wouldn't it possible to generate Ruby/DL
code instead of Ruby C wrapper code using SWIG? This would make it a lot
more easy to use extensions cross-platform without the need for compilation.

Regards,

Peter

I'm no SWIG expert, but I should think it would be possible.

However, my recommendation: hold off on doing anything significant
with Ruby/DL for now. Until Ruby/DL2 comes out, anyway.

- Jamis

 

[Lyle Johnson]

Just thinking out loud, but wouldn't it possible to generate Ruby/DL
code instead of Ruby C wrapper code using SWIG? This would make it a lot
more easy to use extensions cross-platform without the need for compilation.

You might be able to get something to work for some limited cases,
where the code to be called is regular C code. I don't know how you'd
get this to work for C++ code, however (e.g. how does Ruby/DL deal
with name mangling, calling constructors and destructors, templates,
exceptions, ?)

 

[Carl Youngblood]

However, my recommendation: hold off on doing anything significant
with Ruby/DL for now. Until Ruby/DL2 comes out, anyway.

- Jamis

Any word on when that is going to be?

Carl

 

[Jamis Buck]

Any word on when that is going to be?

Carl

Not anytime soon, according to the author. Although you can follow his
progress on RubyForge: http://rubyforge.org/projects/ruby-dl2. The
code's all in CVS.

- Jamis

 

[gabriele renzi]

Lyle Johnson ha scritto:

You might be able to get something to work for some limited cases,
where the code to be called is regular C code. I don't know how you'd
get this to work for C++ code, however (e.g. how does Ruby/DL deal
with name mangling, calling constructors and destructors, templates,
exceptions, ?)

it does not, DL is for C, not c++. IIRC a related note from the author
was that writing a DL for c++ would be damn close to writing a c++
compiler (ok, maybe not this exact words