On Thu, Nov 13, 2025 at 02:45:37PM +0100, Heinrich Schuchardt wrote: > On 11/13/25 14:30, Tom Rini wrote: > > On Thu, Nov 13, 2025 at 07:46:50AM +0100, Heinrich Schuchardt wrote: > > > > > From: Simon Glass <[email protected]> > > > > > > This is an attempt to improve the structure of the FAT code, since it > > > doesn't fully follow the U-Boot conventions: > > > > > > - fat_write.c includes fat.c which is odd > > > - use of __u32 and its ilk > > > - use of typedef > > > - old-style struct comments > > > > > > This series resolves these problems, making it easier to take on other > > > improvements in future. > > > > We don't have an AI policy for the project at this time. My current > > inclination is we should follow: > > https://docs.postmarketos.org/policies-and-processes/development/contributing-and-ai.html > > I cannot see that Simon's work would violate a policy like the one above.
The "please, don't" part. There's no reason to have Claude or any other LLM do basic clean up and refactor work that a human could easily do. > > So I'm inclined to defer this until there's some "other improvements in > > the future", that are both useful > > Could you, please, describe what you dislike about the content of the > changes. Sure. I've seen first hand how hard it is to get current Claude to really actually do what you tell it, and not make subtle changes on its own, even when you tell it to not do that, really. I have no time to review patches that should be simple and if were done by a human would be simple to see what subtle bugs may have been introduced in code that can suffer from both subtle functional bugs and subtle security bugs. > > and hopefully only human developed. > > With your reaction you will probably push contributors to simply not mention > which tools they have used. https://blog.xkeeper.net/thoughts/rants/arguing-against-inevitability/ And if we can't trust contributors to follow the rules, they shouldn't be in the project. -- Tom
signature.asc
Description: PGP signature
