Mega-merges (merge workflow)
Keeping multiple parallel streams of work running, with the main checkout combining them all.
Start
Create a merge commit with multiple parents:
jj new c n x
Mark this merge commit:
jj commit -m '[merge]'
Mark the wip commit:
jj describe -m '[wip]'
All work should happen on this commits (or commits after that).
Squash into one of the workstreams
To squash things back into the correct workstream:
jj squash --into n --keep-emptied
where n is the latest commit of that workstream.
--keep-emptied keeps the wip commit after the merge.
Adding a new workstream
To create a new arm of the merge:
jj new m
where m is the commit that points to the common ancestor, e.g. the main branch.
Then rebase the merge on top:
jj rebase -s orl -d "all:orl-" -d wtm
Rebase orl and its descendants on top of:
- all of
orl's existing parents (the previous merge parents) - and
wtm
Most of this workflow I got from the following sources: