Bug 246364 - science/py-scipy: upgrade to 1.2.3
Summary: science/py-scipy: upgrade to 1.2.3
Status: New
Alias: None
Product: Ports & Packages
Classification: Unclassified
Component: Individual Port(s) (show other bugs)
Version: Latest
Hardware: Any Any
: --- Affects Only Me
Assignee: freebsd-python (Nobody)
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2020-05-10 18:42 UTC by Thierry Thomas
Modified: 2020-05-12 10:17 UTC (History)
0 users

See Also:
bugzilla: maintainer-feedback? (python)


Attachments
Patch to upgrade SciPy to 1.2.3 (2.62 KB, patch)
2020-05-10 18:42 UTC, Thierry Thomas
no flags Details | Diff
Poudriere log. (82.84 KB, application/x-bzip)
2020-05-10 18:44 UTC, Thierry Thomas
no flags Details

Note You need to log in before you can comment on or make changes to this bug.
Description Thierry Thomas freebsd_committer 2020-05-10 18:42:03 UTC
Created attachment 214364 [details]
Patch to upgrade SciPy to 1.2.3

- Upgrade SciPy to 1.2.3
  Release notes at https://github.com/scipy/scipy/releases/tag/v1.2.3

- Register the dependency to openblas (inherited from NumPy)

- Add a test target.
Comment 1 Thierry Thomas freebsd_committer 2020-05-10 18:44:50 UTC
Created attachment 214365 [details]
Poudriere log.
Comment 2 Sunpoet Po-Chuan Hsieh freebsd_committer 2020-05-11 21:03:12 UTC
(In reply to Thierry Thomas from comment #0)
> - Register the dependency to openblas (inherited from NumPy)

It should not be added as it depends on what option is chosen while building py-numpy. For example, a user chose ATLAS may encounter the conflict between atlas and openblas.

from math/py-numpy/Makefile:
OPTIONS_SINGLE_BLASLIB= ATLAS NETLIB OPENBLAS
OPTIONS_DEFAULT=        OPENBLAS SUITESPARSE
Comment 3 Thierry Thomas freebsd_committer 2020-05-12 10:17:40 UTC
(In reply to Sunpoet Po-Chuan Hsieh from comment #2)

This dependency is added to make the package coherent. Since OpenBLAS is the default for NumPy, SciPy should use it.

For ports, this is a bit different: SciPy is smart enough to detect every BLAS package installed on the machine (it will even catch math/blis) and link against them. Of course, if the user builds NumPy with a BLAS library, we can expect that this same library is still installed on the machine when SciPy is built, but this could produce a mess when the dependencies are registered! Such things may happen when ports are built on a non-clean environment, and it is often difficult to prevent them.