[pacman-dev] Removal of bindings
Hi, I don't know why language bindings generation is part of the pacman-lib CVS repository. Any swig-supported (not only perl/python/java) language binding can be generated from alpm.h and a swig interface file. Real language bindings will also ship higher level function/classes. So i would prefer to remove automated bindings generation. Jürgen
On 1/31/07, Jürgen Hötzel <juergen@hoetzel.info> wrote:
I don't know why language bindings generation is part of the pacman-lib CVS repository. Any swig-supported (not only perl/python/java) language binding can be generated from alpm.h and a swig interface file.
That's bothered me for some time as well, but I didn't put too much thought into it. Don't we need the alpm.i conversions, though? Should that be distributed some way?
On 1/31/07, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
On 1/31/07, Jürgen Hötzel <juergen@hoetzel.info> wrote:
I don't know why language bindings generation is part of the pacman-lib CVS repository. Any swig-supported (not only perl/python/java) language binding can be generated from alpm.h and a swig interface file.
That's bothered me for some time as well, but I didn't put too much thought into it. Don't we need the alpm.i conversions, though? Should that be distributed some way?
It doesn't look like there is whole lot in that alpm.i file at the moment, and I'm not sure what those pointer_cast functions are doing. Why don't we make a base-level folder named extra, write up a README file describing the things that are in it, and put things like alpm.i in there? Then we can simplify the building process here once at a time. I think this will also clean up Makefile.am and configure.ac a good amount if we get rid of the bindings always being built. -Dan
On Wed, Jan 31, 2007 at 04:28:39PM -0600, Aaron Griffin wrote:
On 1/31/07, Jürgen Hötzel <juergen@hoetzel.info> wrote:
I don't know why language bindings generation is part of the pacman-lib CVS repository. Any swig-supported (not only perl/python/java) language binding can be generated from alpm.h and a swig interface file.
That's bothered me for some time as well, but I didn't put too much thought into it. Don't we need the alpm.i conversions, though? Should that be distributed some way?
The supplied alpm.i is just a (very simple) example about how to bind c data types. A Real-World binding would use Type-Maps to provide smart bindings. There is no "single" interface file to meet all needs. Jürgen
participants (3)
-
Aaron Griffin
-
Dan McGee
-
Jürgen Hötzel