Re: [PATCH 1/6] revert: Don't remove the sequencer state on error

Ramkumar Ramachandra wrote:

The cherry-pick/ revert machinery now removes the sequencer state when
do_pick_commit returns a non-zero, and when only one instruction is
left in the todo_list.  Since do_pick_commit has a way to distinguish
errors from conflicts using the signed-ness of the return value,
utilize this to ensure that the sequencer state is only removed when
there's a conflict and there is only one instruction left in the
todo_list.

I'm having trouble parsing this.  Is the idea of this one to mitigate
some of the problems with the "remove sequencer state when a conflict
is encountered in the last commit" hack, by suppressing such behavior
when there is an internal error rather than a conflict?  Why bother,
when the behavior is suppressed altogether later in the series?