lhotari commented on PR #24922: URL: https://github.com/apache/pulsar/pull/24922#issuecomment-3471548330
It would be worthwhile to create a [PIP](https://github.com/apache/pulsar/tree/master/pip#pulsar-improvement-proposal-pip) to cover these changes. In this case the reference is currently to #24600 as the motivation. However, there's no clear problem description or goals what is done and why. For reviewers, it's most necessary to understand the problem that the PR intends to solve. That's one role of a PIP document. For example, [PIP-430](https://github.com/apache/pulsar/blob/master/pip/pip-430.md) covered broker cache refactorings and improvements. With an LLM, you can generate most of the PIP document by providing relevant details as context documents to your prompt. The process is somewhat similar to the one described for [PR descriptions with LLMs](https://gist.github.com/lhotari/81f533af4b9ad515e02d96e543c4408b). Context: - GitHub issues as printed as PDFs from the browser or copy-pasted to text files - PR description printed as PDF from the browser - implementation PR patch (append `.patch` to PR url to get the patch file, download and save to .patch.txt file) - pip/README.md - pip/TEMPLATE.md - pip/pip-379.md and pip/pip-430.md as examples The prompt could be such as > I'm writing "PIP-XXX: Optimized in-memory delayed delivery tracker". > Please take a look at the instructions and samples of PIP-379 and PIP-430 and create a pip-XXX.md document that it's a professional Apache Pulsar Improvement Proposal that I could send out for discussion in the Apache Pulsar development community. The expected audience is IT architects and Software Engineers specialized in data streaming. > Return the document as a downloadable markdown file called pip-XXX.md LLM will generate 80% of the PIP document if you already have written a lot of the background material in other discussions. This solution requires a LLM model with sufficient context window size. If the window size is not sufficient, you can use text files instead of printed PDFs for the existing issues and background context discussions. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
