[Mlir-commits] [mlir] [mlir][DLTI] Handle builtin types when combining data layouts (PR #222874)

[Oscar Priego]

Opriego wrote:

> Can you please disclose AI use in the commit message / PR description?
> 
> Also trim the description. For example, we don't need all the section about regression coverage and validation. The description is largely saying things about "what" Which can be inferred from the patch, what's more important in a commit message is the context and "why", Since there isn't much to say here, we don't need to expand into so many paragraphs (looks like AI slop).

I do use AI-assisted tooling to help me navigate the codebase, investigate issues, implement parts of a change, and validate results. It is a tool that helps me contribute more effectively, but it does not replace the engineering judgment, understanding, review, and responsibility required to make the change.

I’d also ask you to be mindful of your tone and wording. Feedback can be given without sounding dismissive, and I think we can be more welcoming to people who are joining the project and trying to contribute.

Learning and navigating a codebase as large and complex as LLVM is not trivial, and meaningful contributions take real effort. I’m completely open to technical feedback and to discussing how a change should be presented, but I’d appreciate that feedback being expressed constructively and respectfully.

Regarding the PR description, I’m happy to remove actual repetition, but I prefer to keep useful information about scope, regression coverage, and validation. I use the PR as part of the engineering record of what was exercised and what the change is intended to cover.

https://github.com/llvm/llvm-project/pull/222874