[PATCH] rebase: split _checkobsrebase into two parts for better wrapping

[Pierre-Yves David]

On 05/30/2016 07:31 PM, Kostia Balytskyi wrote:
> # HG changeset patch
> # User Kostia Balytskyi <ikostia at fb.com>
> # Date 1464629336 -3600
> #      Mon May 30 18:28:56 2016 +0100
> # Node ID 330fa585c8b9a272f51e6b82a8f65d9b01e0207e
> # Parent  318534bb5dfd37b1c95f10ba90b73c5ce562713f
> rebase: split _checkobsrebase into two parts for better wrapping
> 
> Currently _checkobsrebase does two checks: one aborts if divergence is going
> to be caused by the rebase operation (and this can be turned off by a config),
> another aborts if all changesets have successors in the destination (and this
> cannot be turned off). I have suspicions that some rebase users would like it
> not to abort when there's nothing to rebase, but rather to exit with code 0.
> Splitting _checkobsrebase is a step in this direction.
> 
> This is unrelated to the other refactoring series.

I actually lean toward have that alternate behavior as the default, I
can't remember the discussion with Laurent that led to the current
default (if we had one), but the current abort feel strange to me and I
would rather see the rebase proceed successfully (it will have to update
the working copy parent and related bookmarks)

I've CC Laurent who might have valuable insight on why the current
behavior was picked.

-- 
Pierre-Yves David