Re: [PATCH 3/5] pull: handle conflicting rebase/merge options via last option wins

Elijah Newren <newren@gmail.com> writes:

Yeah, it's about time somebody made an exhaustive table of cases ;-)

I appreciate it very much. Without such a table, it is hard to have discussion and keep track of which part is and which part is not what people agree to.

Below, I rearranged some rows to group cases I think belong together to make commenting on them easier, but I didn't remove or add any rows.

I think these three make sense. As you answered to a question in another thread, I agree that --ff-only from the command line should fail instead of letting rebase (or merge for that matter) go ahead when their history is not a descendant of ours and it is a bug if it does not.

OK.

OK.

I think the second one deserves an explanation. The rationale for ignoring --ff-only is because the act of giving an explicit "--rebase" or "--no-rebase" from the command line, when the configured default is "I expect to have no development on my own here, and only want to follow along", is a sign enough that the user does not want to follow along in this particular invocation of "pull". And the rationale for the entire thing to become --ff is only because between --ff and --no-ff, the former is the default.

About the second one, I would understand it if it became "merge --ff-only", too. That is more trivially explained. I however suspect that it would be less useful, but that is open to discussion.

All three others I think are not controversial.

Again the same reasoning applies to the second case that defeats the configured pull.ff=only setting. pull.ff=unset and pull.ff=true would be handled the same, so only the pull.ff=false might deserve an explanation. I not think there needs anything special, as there is no "I want to do something special to mark the fact that we tried to integrate their work" special case needed even when their history is descendant of ours in "rebase" workflow, where we replay what we did on top of theirs (in other words, in "merge" workflow, the shape of history left with --no-ff would be different from --ff; in "rebase", there is no difference---either you had your own development before you ran "pull --rebase" and they got replayed on top of their history, or you had nothing before you ran "pull --rebase" and you did your development after that on top of their history, you'd end up with histories of the same shape. This would not happen with "merge" workflow and that is why --no-ff may matter).

So, I agree with the above 4 (and the other 4 for --no-rebase) entries.

OK.

OK.  We do specified rebase, allowing fast-forward.

pull.rebase=<something> combined with --no-ff could mean "I do expect I have some development of my own, and please stop me if I don't" and instead of rebasing nothing, error out without even repointing our branch tip to their history when it fast forwards, but I suspect that it would be less useful. Again, that (and anything else I said here) is open to discussion.

Thanks.