Re: Alternative for scipy.sparse.sparsetools for use from outside of scipy

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

Re: Alternative for scipy.sparse.sparsetools for use from outside of scipy

Lev Konstantinovskiy
What is the danger of using sparsetools from outside of scipy?

I found the Mar 2014 PR that made it internal but would be grateful to
know the reasoning. "MAINT: wrap sparsetools manually instead via
SWIG#3440" https://github.com/scipy/scipy/pull/3440

Thanks
Lev
_______________________________________________
SciPy-User mailing list
[hidden email]
https://mail.scipy.org/mailman/listinfo/scipy-user
Reply | Threaded
Open this post in threaded view
|

Re: Alternative for scipy.sparse.sparsetools for use from outside of scipy

Pauli Virtanen-3
Lev Konstantinovskiy <lev.konst <at> gmail.com> writes:
> What is the danger of using sparsetools from outside of scipy?

It is a private API. Changes can be made to it without needing
to go through deprecation cycles.

The boundaries of what is private and what was not historically
very clear in Scipy (and in Python in general). Clearly, it does
not make sense to keep all helper routines as public.

https://docs.scipy.org/doc/scipy-dev/reference/api.html#api-definition

The case of scipy.sparse is that it will likely need to go through
some restructuring in the future, partly because of the need for
more array-like sparse matrices, and also partly to improve
performance in places where the half-Python half-C++ approach
has too large costs.

--
Pauli Virtanen


_______________________________________________
SciPy-User mailing list
[hidden email]
https://mail.scipy.org/mailman/listinfo/scipy-user