xakep8 wrote: > @xakep8 excuse my possible rudeness, but how did you trace the entire feature > and wrote a comment that consists out of about 1500 symbols in just 18 > minutes? It feels like some parts of your comment might be a pure LLM output. > Can you make sure to always write any comments by yourself and make sure to > disclose AI usage in your contributions as per our > [policy](https://llvm.org/docs/AIToolPolicy.html)? > > > I'll split this into the CIR cleanup > > I'm not entirely sure what cleanup do you mean? This PR seems to be adding > new functionality rather then cleanup.
Yes some parts of that comment are actually AI generated but they are not from the 18 mins that I had, they are from the analysis I did before this whole PR and it was just to find files and to understand what is flowing through where and yes I have used some parts of that analysis document/notes to question and reply to the comment you added, I had actually traced this before even raising this PR and I was actually question my AI if raising a warning for reachable unsupported code is a good thing or not, in fact it told me that we should be throwing and error and those are not my thoughts I did not really trust that and wanted to go through everything myself to even come to a conclusion. All the code I have submitted for this PR came from a lot of back and forth understanding from AI but they were not specifically generated I'd say, I'll check the document and add a disclosure if the analysis I do with AI comes into a disclosure category. That said I'm really sorry that you had to flag me for this, I just thought that the explanation I had in the document/notes was better and more specific to someone who has worked with the code because knowing that from my thoughts they could flag out if I'm wrong in my approach somewhere. Also since the type mismatch was in CIR and not in lowering that's why I called it a cleanup, I understand that this is actually a new functionality being added. https://github.com/llvm/llvm-project/pull/227595 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
