Repository navigation
The Recommended Random Number Generation In NumPy #19670
Description
Problem
Matplotlib documentation still uses old method of random number generation in NumPy. I think we can update the code to use newer method. Newer method was intropduced in NumPy 1.17 and recommended by NumPy over the older method (even though the older method still works)
- I found histogram example can be updated with newer code.
Suggested Improvement
- We can change the code to shown in this GitHub gist
- Entire documenation regarding recommended method is in NumPy docs
- Here is a visualization/output from the code in gist above:
Matplotlib version
- Operating system: Arch Linux
- Matplotlib version: 3.3.4 (matplotlib.version)
- Python version: 3.9.2 (platform.python_version())
- Matplotlib documentation version: Doc version v3.3.4 (as seen at the bottom of the histogram page )
Reacted by Antony Lee and Gregor Simm
Activity
ria-19 commentedon Mar 9, 2021
Hi, I am a beginner in the open-source world, can I work on this issue?
jklymak commentedon Mar 9, 2021
Anyone can work on an issue if there is no existing PR. If there is a PR it's nice to ask the author first.
ria-19 commentedon Mar 9, 2021
Okay, thank you.
rajpratyush commentedon Mar 14, 2021
@jklymak Can you tell me where the file to be changed are located?
story645 commentedon Mar 14, 2021
@rajpratyush unless someone happens to know off hand, it'll take them about as long to dig into the docs to find it as it would take you. One method though of finding the right file is running 'git grep' on your local version of the repo w/ some words that are specific to the example.
ArnuldOnData commentedon Mar 14, 2021
@rajpratyush Whole of the process and how-to, for contributing your code, is explained here:
https://matplotlib.org/stable/devel/index.html
rajpratyush commentedon Mar 14, 2021
Sure @story645 Lemme look for it. Thanks for assistance
tacaswell commentedon Mar 18, 2021
As discussed in #19706 , we have decided to not change all of the use of random number generators in our examples. Thank you for the suggestion @ArnuldOnData , but we are going to close this with no action.
It may be reasonable to re-consider this is a few years when pre 1.17 versions of numpy are few and far between or if numpy starts actively warning when the old API is used.