Hey Apache Training team, it is time to close a pull request I opened during my warm-up phase some weeks ago. I mean https://github.com/apache/incubator-training/pull/87.
One comment was related to the usage of LLMs and how to handle their output. In this source we find ASF roles and recommendation related to LLMs in ASF projects: https://www.apache.org/legal/generative-tooling.html *How does this relate to the proposed tool I created?* Let me start with an explanation of the purpose of this tool. Although a huge number of new GPT based online tools emerge every day, the intention of writing such a document inspection self on my own was purely driven by the desire to understand the requirements for such a tool from the ground up. *With this in mind, the expected outcome of this subproject is as follows:* [ 1 ] => Given a cloned ASF code base, *as a Trainer* *I want to be able to "analyse" this code base regarding existing documentation.* We all know the test coverage reports generated with Maven. I think about something comparable for documentation. But this vague idea needs some better shape, hence I share it with the community so that we can grow it together. [image: image.png] What do you think about a "training readyness tag" in ASF projects, after evaluating the documentation - I am sure we can automate this a lot in the near future. [ 2 ] => Providing a pointer to specific artefacts in this repository, *as a Trainer I want to be able to answer questions of my student* to the particular code base. [ 3 ] => Specifically, focussing on the project README.md file, *as a Trainer, I want to understand the documentation status of the project*. How easy or complicated would it be to get onboarded as a greenhorn, a solid developer, or even as an expert in a domain with not so deep coding experience? In order to get the pull request closed I will focus on [3] and leave the other two points to you. Please share your thoughts. Towards an implementation, I think there are so many options to be considered, e.g., I selected Python for implementing the initial tool because I had a working example, but it could be, that this gets thrown away later. I suggest, that we focus on the questions behind particular tools and define the demand from a trainers perspective. And here is an important aspect to be aware of, we should expect, that the way how training is delivered will change faster than we develop procedures or content. This is a great challenge and also a great opportunity to support communities. What are your thoughts? Cheers, Mirko -- Dr. rer. nat. Mirko Kämpf Müchelner Str. 23 06259 Frankleben
