Jeff King <peff@xxxxxxxx> writes: > I think 51441e64 is essentially a revert of 2386535511 (attr: read > attributes from HEAD when bare repo, 2023-10-13). I don't know how you > prepared it, but I'd probably have started with "cherry-pick -n". But > that wouldn't help, because the documentation didn't come until after > that in 9f9c40cf34 (attr: add attr.tree for setting the treeish to read > attributes from, 2023-10-13). "revert -m 1" followed by "commit --amend" might have worked well in this case to get rid of the code that came from one and doc update that came from the other in a two patch series, but in general, that would be too much noise to wade through in general. > Not that it really matters much now, but always just curious about how > we can avoid missing stuff like this next time. The series first did "HEAD tree is used in bare" without doc, and followed up with "configuration can be used to name any tree" with doc that mentions the behaviour of the first step as a special case of default value for the configuration variable. The only way it could have been made easier to spot is to introduce the variable with documentation first, and then do the "bare repo uses HEAD as the default" thing on top.
