[Numpy-discussion] performance matrix multiplication vs. matlab
Mon Jan 18 13:39:26 CST 2010
On Mon, Jan 18, 2010 at 13:34, Vicente Sole <email@example.com> wrote:
> You are taking point 4.d)0 while I am taking 4.d)1:
> 1) Use a suitable shared library mechanism for linking with the
> Library. A suitable mechanism is one that (a) uses at run time a copy
> of the Library already present on the user's computer system, and (b)
> will operate properly with a modified version of the Library that is
> interface-compatible with the Linked Version.
> If you are using the library as a shared library (what you do most of
> the times in Python), you are quite free.
numpy would not be using Eigen2 as a shared library. It is true that
numpy would act as a shared library with respect to some downstream
application, but incorporating Eigen2 into numpy would make those
numpy binaries be effectively under the LGPL license with respect to
the downstream application.
> In any case, it seems I am not the only one seeing it like that:
> The key point is if you use the library "as is" or you have modified it.
With respect to numpy and the way that Eigen2 was proposed as being
used, no, it is not the key point. We will not incorporate Eigen2 code
into numpy, particularly not as the default linear algebra
implementation, because we wish to keep numpy's source as being only
BSD. This is a policy decision of the numpy team, not a legal
"I have come to believe that the whole world is an enigma, a harmless
enigma that is made terrible by our own mad attempt to interpret it as
though it had an underlying truth."
-- Umberto Eco
More information about the NumPy-Discussion