On Tue, Dec 12, 2023 at 05:08:34AM -0800, Christoph Hellwig wrote: > On Tue, Dec 12, 2023 at 05:09:21PM +0900, Masahiro Yamada wrote: > > Unlike .gitignore, this feature is opt-in rather than enforced. > > > > To use this, you need to run > > > > 'git config diff.orderFile scripts/git.orderFile' > > > > or > > > > 'git diff -C scripts/git.orderFile' > > Oh, ok. That greatly reduces my concern. Yes, it's an opt-in, so no user should be directly impacted. > > > > > Indeed, the file order is subjective, leaving > > us a question "do we need it in upstream"? The main idea is patch generation. This file's order is supposed to be the best order for reading a raw patch and understanding the code changes. > > > > At least, it is harmless for people who have no interest. > > .. but this is still a good question. I'm not really sure there is > much of a need for it, but as long as it doesn't harm everyone else > I'm at least neutral on it. diff.orderfile was introduced in git to help order the git diff, and thus the patch generation, in a way that it's easier to understand what the commit / patch intends on doing. Take this example introducing a feature foo, you should see: - Documentation on foo, if introduced - How is foo enabled in build system, if needed - The types / stucts / fields introduced by foo, if any - The interface for using foo, if any - The actual foo implementation. Of course the actual order is open to discussion, and I encourage everyone to suggest any other items or order. Thanks! Leo
