On Tue, Jun 19, 2012 at 12:07:27PM -0500, Steve Huston wrote:
> Collecting the bindings seems like a good idea, but how do you envision
> building the bindings? You mentioned breaking the cmake dependencies. Are
> you planning to break the dependency on the cpp directory's
> CMakeLists.txt, or on cmake all together? What triggers the need to build
> the bindings when the cpp dirs change?

We wouldn't break away from cmake altogether. By that I meant that, if
we started versioning the SWIG code generated currently, then we would
not have a direct dependency from, say, the Perl bindings onto the C++
code. We could build it completely separately from C++.

And like I mentioned in my email to Andrew, I'm not sure I like that
idea. I would rather have the C++ code be its own single cmake project.
Then have either the individual language bindings each be single
projects that then depend (via cmake) on the C++ projects.

So that way I could do:

cmake ../qpid/cpp                # to build the C++ code
cmake ../qpid/bindings/qpid/perl # to build the Perl bindings, which
                                 # would in turn build C++ as needed
cmake ../qpid/bindings/qpid/ruby # ibid, for Ruby
cmake ../qpid/bindings           # build C++ then all language bindings
cmake ../qpid                    # build the world


-- 
Darryl L. Pierce, Sr. Software Engineer @ Red Hat, Inc.
Delivering value year after year.
Red Hat ranks #1 in value among software vendors.
http://www.redhat.com/promo/vendor/

Attachment: pgp3f08OSVNPC.pgp
Description: PGP signature

Reply via email to