Hi, The way we plan is to have the package compyte[1] that is already used by PyCUDA and PyOpenCL contain the new code. This package will be a dependency of PyCUDA, PyOpenCL, Theano and any other people that want to use the same interface directly.
I think it is not complicated, but I never did a distro package. Is that ok for you? We create a new mailing list to discuss the creation of the gpu ndarray[2] So this thread will be moved there. We encourage all people that want to have a look or participate in this, to subscribe. Early choice will have important long term impact. thanks Frédéric Bastien [1] https://github.com/inducer/compyte [2] http://lists.tiker.net/listinfo/gpundarray On Fri, May 13, 2011 at 10:49 AM, Tomasz Rybak <bogom...@post.pl> wrote: > I am making Debian packages for both PyCUDA and PyOpenCL > and I would like to avoid having complicated set of dependencies. > In other words, I would not like to have situation when > GPUCArray depends on PyCUDA which in turn depend on GPUCArray; > this makes building package really hard. > > Best regards. > > -- > Tomasz Rybak <bogom...@post.pl> GPG/PGP key ID: 2AD5 9860 > Fingerprint A481 824E 7DD3 9C0E C40A 488E C654 FB33 2AD5 9860 > http://member.acm.org/~tomaszrybak > _______________________________________________ PyCUDA mailing list PyCUDA@tiker.net http://lists.tiker.net/listinfo/pycuda