[PATCH] rebase: add flag to commit to insert commit in stack

[Durham Goode]

On 11/25/15 3:04 PM, cdelahousse at fb.com wrote:
> # HG changeset patch
> # User Laurent Charignon <lcharignon at fb.com>
> # Date 1448481218 28800
> #      Wed Nov 25 11:53:38 2015 -0800
> # Node ID e08ef9c3e1409b9864e14f5b66ddd3821b4d6115
> # Parent  397baacd694d0f8343888b64c200bfa926cdd893
> rebase: add flag to commit to insert commit in stack
>
> This patch adds a flag to insert a commit within an existing stack without
> forcing the user to do a separate rebase step. From the call to commit --insert,
> the tool flag is passed into our call to rebase in case the user specifies it.
>
> A new test file is created because as there was no existing suitable place to
> test the functionality.
>
The test file appears to be missing?

If there are conflicts is it obvious to the user that they need to run 
hg rebase --continue?

Also, what happens if I do 'hg commit --amend --insert'?

If we're going to allow this kind of flag, it opens the doors for a lot 
of other interesting flags (hg commit --amend --rebase ; for amending 
with a rebase. hg commit --amend --into X ; for amending pending changes 
into a commit further down the stack).  Do you see the '--insert' 
grammar making sense anywhere else (hg rebase -r X -d Y --insert ; to 
insert X between Y and it's desendants)?  Just seems like we'd want to 
strive for a consistent ux around these operations, and if we get it 
right it could be very powerful.