Hi Karli, Karl Rupp <r...@iue.tuwien.ac.at> writes: > I was currently in the process of running Philippe's autotuner. So it > turned out that the installation process aborted because my stock > installation of NumPy (1.6.1) is outdated. My machine runs a Linux Mint > Maya, which is built on top of Ubuntu 12.04.
Try changing lines 225 and 226 in setup.py so that the inequality is true for numpy 1.6. I haven't tried building against that, but it may well work. > Is there any chance we can also support such 'older' setups, > particularly as this is an LTS release and supported for a few more years? It depends on if boost-numpy supports older numpy, and if not, how much work it will be to modify it :) > Note: I'm not yet talking about Enterprise OSes like RedHat 5 (CentOS > 5), which are still in use at companies... We should make reasonable > efforts to also support such machines, not every user has the privilege > of root access and the time and will to manually install the latest > versions. It's easy to install Python packages locally. Just pass '--user' to `setup.py install`, or to `pip install`, whichever you are using. Though indeed, we should keep things as seamless as possible. Note that if you were running `pip install pyviennacl` -- as I expect most users to do -- and you didn't have a modern numpy installed, then pip would install one for you as a consequence of the dependency. Cheers, Toby ------------------------------------------------------------------------------ Want fast and easy access to all the code in your enterprise? Index and search up to 200,000 lines of code with a free copy of Black Duck Code Sight - the same software that powers the world's largest code search on Ohloh, the Black Duck Open Hub! Try it now. http://p.sf.net/sfu/bds _______________________________________________ ViennaCL-devel mailing list ViennaCL-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/viennacl-devel
