[zuprau]

In that case I also see "large" PRs, but I point the reviewer to "file X and then 4999 copies of file Y." When doing large no-change refactors, I submit standalone PRs and merge them quickly because no one will review 1k identical changed lines (e.g. if I change the location of /lib/ imported 600 times in the project)

[tsimionescu]

What if the refactoring makes the code harder to understand when looked at in isolation, but is necessary for the rest of the feature to work? Why submit it in a separate PR without the context of why it's necessary?