Package 'python3_numpy' not working on nightly build (hrev52992)

Development
1

Numpy is one of those essential general usage packages for people working with Python. However, in my system, the Python 3 version of Numpy seems to be improperly configured, despite the fact that it was instaled using HaikuDepot. The Python 2 version imports normally, however.

At first I thought that I could have messed up this 64 bit Haiku installation, so I rebooted into my 32bit Haiku partition to see if it worked there, but I found the same issue.

It’s really sad, because there are a lot of useful Python packages that depend on Numpy. Any hope to get this fixed?

By the way, Python 3.7 was released 9 months ago, and it’s now at v.3.7.2 (v.3.7.3 expected later this month, and v.3.8.0 expected to october). Compared with Python 3.6, Python 3.7 is much faster in some operations, and it also includes a few useful new features. Does anyone know if there is already a port coming soon through the pipeline?

================================================

~> pkgman search numpy
Status  Name                  Description                                                               
--------------------------------------------------------------------------------------------------------
S       python3_numpy         Array processing for numbers, strings, records, and objects               
    python3_numpy_source  Array processing for numbers, strings, records, and objects (source files)
S       python_numpy          Array processing for numbers, strings, records, and objects               
    python_numpy_source   Array processing for numbers, strings, records, and objects (source files)
~> python3
Python 3.6.7 (default, Dec 18 2018, 19:22:41) 
[GCC 7.3.0] on haiku1
Type "help", "copyright", "credits" or "license" for more information.
>>> import numpy as np
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ModuleNotFoundError: No module named 'numpy'
>>>
2

Please use the HaikuPorts Issues for reporting bugs with 3rdparty software, the forum is not the right place for that.

In the meantime:


and

and

1 Like
3

Thank you very much.

4

Would you test my recipe? I was able to import it, but i have no idea how to use it actually, so my testcase is pretty limited.
You can build it with haikuporter after pulling my PR.

5

I would gladly do it, but I think it will be my first time with haikuports recipes and manual builds on haiku. Which step should I follow? Is there any tutorial?

6

The installation and git cloning of haikuports (beware, “lot” of data, network traffic and space required).
Installation of Haikuporter

Some info on recipes and so:
Introduction to Haikuporter 1

Built succesfully in latest 64b nightly (clean state install) and imported ok. All examples that worked over arrays run OK on python 3.6 :+1:
More expert usage of numpy may check it better.

2 Likes