Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

choice of a name for an openblas wheel #3

Open
carlkl opened this issue Feb 8, 2016 · 3 comments
Open

choice of a name for an openblas wheel #3

carlkl opened this issue Feb 8, 2016 · 3 comments
Labels
Milestone

Comments

@carlkl
Copy link
Contributor

carlkl commented Feb 8, 2016

I intend to deploy OpenBLAS as a wheel, that installs all files into the mingwpy folder. For the windows/mingwpy case mingwpy is located in <pythonroot\share\mingwpy>, hence OpenBLAS as well as other libraries would be placed into <pythonroot\share\mingwpy\opt>.

I want to recall @njs suggestion to name wheels with library development files clib_<libraryname>##.whl, i.e. clib_openblas-0.1.15-cp34-none-win_amd64.whl. Is this the right approach?

@carlkl carlkl added the question label Feb 8, 2016
@matthew-brett
Copy link
Member

Sounds reasonable to me.

@ogrisel
Copy link
Contributor

ogrisel commented Feb 21, 2016

+1 as well.

@carlkl
Copy link
Contributor Author

carlkl commented Feb 22, 2016

As a remark to the hangout yesterday:

A somewhat rushed written RFC: let the BLAS/Lapack be interchangable by the user

I near future the idea to use a generic name for a DLL containing BLAS/Lapack routines could be tested, i.e. name it clib_blaslapack.dll. This DLL should export all standard BLAS and Lapack routines.

To would allow the user to change the underlying BLAS/Lapack machinery according to its needs.

To make a change of the implemenation happen a DLL provided by an extra package has to be loaded into the process space beforehand import of numpy. This DLL could be loaded into the process space with the help of ctypes. The selection of the desired implementation could be done with the setting of a environment variable or with a small helper function.

numpy itself could be deployed with a default implementation, say ATLAS, so changing this is optional and should be controlled by the user.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
None yet
Development

No branches or pull requests

3 participants