Searching for a good issue to work on

[Johannes Erwerle]

Hello!

I am a student studying computer science at the University of Stuttgart
in Germany. We have a "Simulation Software Engineering" course, which is
basically a "How to contribute to open source software" course with a
focus on simulation and research software. You can find further details
about that course on the course homepage:
https://simulation-software-engineering.github.io/homepage/

Part of the course is to actually contribute something to a software
project and I chose networkx because I am often working with graph
datastructures and in this way I can contribute something to a topic
that I am interested in.

However finding a good issue to work on as someone who does not have
deeper insights into the code and the project is quite hard, especially
when most new issues find someone that completes them very quickly.

Some digging on my end resulted in two issue that I think are interesting.

First https://github.com/networkx/networkx/issues/4885 which shows that
there is an issue with the Fruchterman-Reingold graph layout algorithm.
I am familiar with this algorithm and the force-directed graph drawing
algorithms in general and the original author of this issue might have
other priorities in life a the moment. So I could work on that.

An other open issue is https://github.com/networkx/networkx/issues/3997
Here the code is already mostly done, what remains is primarily
integration into the networkx code base, documentation, testing and
further benchmarking.

My question at the maintainers is now: Are those issues something worth
working on? Or do you have other issues that could benefit from me
working on them?

Greetings
Johannes

[Eric BEAUSSART]

Hello Johannes Erwerle !
I know a very special subject ... Knots on trajectory ...
Not exactly "Seamen's knots ...
In "States" Spaces and their duals "Transitions" Spaces, usually
are some ""Strange Attractors" ...
Just read Raymond Poincaré, René Thom, ... and see the pictures ...
Specially 'Deux noeuds" ... Two knots ... showing
two "helicity" ...
Friendly ...

envoyé : 19 janvier 2023 à 14:17
de : Johannes Erwerle <jo+ne...@swagspace.org>
à : networkx-discuss <networkx...@googlegroups.com>
objet : [networkx-discuss] Searching for a good issue to work on

--
You received this message because you are subscribed to the Google Groups "networkx-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to networkx-discu...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/networkx-discuss/464b9e3d-d316-dec1-e3e6-ecaac30b7501%40swagspace.org.

[Dan Schult]

Welcome Johannes,

Thanks for your interest in NetworkX.

I think either of the issues you have proposed are good for an in depth project that someone familiar with the graph theory side can contribute in a few months of class oriented time.  It may take longer for review than for writing, but I think those issues are motivating enough to warrant review energy.  I think either issue would be useful. I guess if I had to pick one it would be FR since you have some experience with that, and the issue is with the boundary conditions more than the force part of the alg. The clustering coefficient issue is good too -- perhaps more focused (less open ended) and might tie in with other packages such as scipy.sparse for matrix manipulation. It would be good to try to mimic the style and API of similar code within NetworkX -- and you should read the contributor and developer docs. Your class will help with some of the git and tool usage -- maybe even pytest. But as you come across problems and questions, open what you've got as a PR (as a DRAFT-PR if you want to let others know that it's not quite ready) and then we can answer questions, make suggestions, etc. The github conversations are a good place to discuss code choices.

Best,

Dan

[Johannes Erwerle]

Hi Dan,

thanks for the quick response.

On 19/01/2023 17.52, Dan Schult wrote:
> It may take longer for review than for writing

I guessed that already when taking a look at the large amount of open PRs ;)

> I think either issue would be useful.

That sounds good.

> and you should read the contributor and developer docs.

I did that already.

> Your class will help with some of the git and tool usage --
> maybe even pytest.

Yeah, the course also contains sessions regarding testing and pytest,
but I am already very familiar with python and the surrounding ecosystem.

I will make my selection at the beginning of the next week, report my
decision in the corresponding issue and start working on it.

Greetings
Johannes