High-level diagrams for pypdf

[ivanmilevtues]

Hi all,
I am Ivan. I love and use OSS all time. However when it comes to contribution I usually get hyped and then intimidated by the volume of these projects. Thus decided to try and help people get to know codebases faster, to me one of the best ways to start working on something is by getting a high-level understanding of the flow, and then deepening on the parts which interest you further. This is why I think diagrams are a great way to start. Let me know what is your take on this!

I am working on generating always up-to-date high-level diagrams for onboarding and generated one for pypdf: https://github.com/CodeBoarding/GeneratedOnBoardings/blob/main/gpt_engineer/on_boarding.md . I would love to hear any feedback on it!

If you want to get in touch with me in private do not hesitate to do so!

[stefan6419846]

Thanks for your interest in pypdf.

As far as I can see, these summaries are generated by LLMs and thus possibly prone to errors. While some possible contributors might find this helpful, I do not really think so. As far as I can tell from having a short look at the generated documents, these only scratch the surface sometimes, while they explain private APIs in other cases which can change at any time and yield the impression that they are especially relevant.

By the way: PyPDF2 is EOL for quite some time and should be replaced by the maintained pypdf.

[ivanmilevtues]

Firstly thank you for the reply!

The content is generated by leveraging Static Analysis and LLM's.

As the diagram is aimed at new people, do you think that more in-depth knowledge is needed to get started. And if so how do you think this will happen i.e. there are more depth levels of the diagram (until we go to pretty much function calls) or the description have to be more details i.e. not just a small paragraph?

Also thanks for the clarification on the EOL for pypdf.