Thanks for your reply. I have opened an issue on github with a simplified example of a travis build that generated linux wheels: https://github.com/inducer/pyopencl/issues/263
Perhaps this could be adapted/integrated into your gitlab-ci pipelines(?) Cheers, Gonçalo On Fri, Jan 25, 2019 at 2:36 PM Andreas Kloeckner <li...@informa.tiker.net> wrote: > Goncalo Morgado <goncalo.ma...@gmail.com> writes: > > > This could be an issue with pip indeed. I believe this problem could be > > workaround if pyopencl provided binaries (wheels) in PyPI. > > > > A related issue is that running `pip install -r requirements.txt` on a > file > > with the following dependencies: > > > > # requirements.txt > > numpy==1.16.0 > > pyopencl==2018.2.2 > > > > will fail, which can be confusing. > > Prompted by > > https://github.com/inducer/pycuda/pull/195 > > I tried to make 'pip install' just work for PyCUDA and PyOpenCL, using > essentially equivalent machinery: > > https://gitlab.tiker.net/inducer/pyopencl/merge_requests/67 > > According to the comments in the issue log, what I did seems to have > worked for PyCUDA, but I don't really know what (aside from the pybind > dependency) is different for PyOpenCL. I'd be happy to consider > patches/PRs that make this work properly, but I myself don't really have > any further ideas or engineering time to devote to this. > > Andreas > _______________________________________________ > PyOpenCL mailing list -- pyopencl@tiker.net > To unsubscribe send an email to pyopencl-le...@tiker.net >
_______________________________________________ PyOpenCL mailing list -- pyopencl@tiker.net To unsubscribe send an email to pyopencl-le...@tiker.net
