Re: git-rebase produces incorrect output

[Junio C Hamano <gitster@pobox.com>]

Elijah Newren <newren@gmail.com> writes:

> Sadly, I tried to force this with git rebase, but -C5 only affected
> the apply side and there's no option to pass to rebase to pass through
> -U5 to the diff logic.  Also, although there is a diff.context config
> option, git-am ignores it (Note that git_am_config() does not directly
> check that value and it calls git_default_config(), not
> git_diff_ui_config() or even git_diff_basic_config()).

Not essential but puzzled.  The context applies to the generation
side, not acceptance side, no?  IOW, I suspect that you are talking
about "git format-patch" that sits on the upstream side of the pipe
that feeds "git am".

> So, to summarize here:
>   * you have a case where the default 3 lines of context mess stuff
> up; but rebase --merge works great
>   * am doesn't have a -U option, and ignores the diff.context setting,
> making it impossible to force the am backend to work on your case
> and also:

I do not think it is super hard to teach "git rebase" to pass
backend specific options so that "git rebase--am" can be told to
work with wider context (which will reduce the risk of ambiguous
patch like this example, trading the increased risk of unnecessary
conflicts; it is a good trade-off most of the time for added safety,
as nobody wants a system that produces a wrong result silently and
quickly).

Having said that,

>   * rebase doesn't have an option to use the merge/interactive backend
> by default (nor an --am option to override it)

I think addition of rebase.backend would be a good first step for
eventually flipping the default, which by the way I have no trouble
with.

> Maybe we should just switch the default, for everyone?  (And provide
> an --am option to override it and a config setting to get the old
> default?)

Yes, that would be a sensible second step.  I actually think a
longer term goal is to deprecate the am backend.  It was invented
first and then kept to be the default backend for a long time
because the merge based backend historically has been noticeably
slow (it was expected to be---it was essentially a shell script that
run cherry-pick repeatedly in a loop).  In some future, it would
outlive its usefulness, and that I think that that future is just
around the corner.