This was the old behavior and produced a lot of headache when the script was restarted on an older branch and then used an older version of the script.

Any particular use case this would solve? You can e.g. change the --work-dir to something else if the default is problematic (the /tmp location is bad due to auto-cleanups).

We want to mostly get rid of this script anyways for most cases, CC hablich for the issue about this.

i haven't figured out yet why, but often merge_to_branch complains about the checkout in /tmp being broken, so I have to nuke it. And then it takes ages to create a new one

we could use git drover :)

You should let --work-dir point to another location, e.g. in your home. The /tmp folders are auto-cleaned every few days and this results in broken git checkouts in there.

We could also implement a feature that chooses the work-dir based on some configuration file, or simply change the default to something non-tmp.

Re 3: Sure! hablich can point to the issue about this.

Re 1: I don't think this should be the default behavior for V8:
- Our checkout does not take forever like Chrome's
- Messes with your own V8 checkout which might have important branches in it
- Can already be overwritten via --work-dir

Cherry-picking instead of squashing is a different story. We could start cherry-picking one CL merges instead of squashing them and thus benefit from the better merge algorithm.

git drover also doesn't mess with your checkout

You can also use git cherry-pick today. Using merge-to-branch gives you the extra features:
1. Find extra ports to merge.
2. Squash multiple commits.
3. Tag the commit.

Feature 3 is not relevant as we have an auto-tag bot. 1 and 2 might not apply in many cases.

Pri-3 with no updates in 180+ days that are available and have no owner == Archived. Feel free to update status, add context as to why and move this along.