I’m working on a documentation issue, which is mentioned in issues #5459.
While reading the existing docstring, I encountered some ambiguity, such as:
It is sometimes referred to as
default 0, and other times as
Defaults to 0.
Some parameters are labelled
float, whereas others are labelled
tensor like of float. Some are
optional, while others are not. Some are labelled
numeric, while others are
numeric or np.ndarray or aesara.tensor, or
I’m new to this, so I was wondering if there are any prerequisites or lessons I should watch in order to comprehend the details of this document.
Welcome and thanks so much for your help! I will let @OriolAbril and/or @almostmeenal speak to the desired way to harmonize/unify some of these.
The goal of the issue is precisely to unify all of these. You should read Your first contribution to PyMC — PyMC-Data Umbrella Sprint 0.1 documentation (link also in the “Details” section of the issue) which explains exactly all of these.
Thanks to the sprint we have even done a couple of minor improvements more like
tensor_like of <type> that I will add to the guide as soon as I have time.
Thank you very much, @OriolAbril.
I watched your video Contributing to PyMC Documentation (Oriol Abril Pla) and now I’m going over the materials again, hoping that this time it will clear my worries.
Note that my webinar is not about docstring updates, it is related to Update example gallery style and formatting · Issue #5460 · pymc-devs/pymc · GitHub. There is also a video about the docstring updates (embedded in the guide I linked above) by Reshama. There were however some questions at the end of my webinar that were related to docstring updates and might be illustrative in this case too.
Hi @OriolAbril, Can I update all the docsting of a file (like: continuous.py) at once and create a single pull request?
You can, but I would recommend around 5 docstrings or so at the same time. The bigger the PR the longer it takes to review, and given how the whole codebase is in active development, the easier for merge conflicts to appear which can be tricky and for me are always annoying to solve. I do understand that a PR per docstring can also be tedious and feel slower than necessary which is why I suggest this half way option of a handful of docstrings per PR
Sure, I will do the same.
I believe the docstring for Continuous has been finished.
Can I begin working for any of the other distributions listed in the API Reference?
Or do you have a specific sequence in mind?
I have no specific order in mind, and I think there is no order to go about modules in pymc either. Feel free to choose whatever docstrings you are interested in, the ones that are related to your gsoc proposal…