Hilary James Oliver

@oliver-sanders - I think there's a small error in the [proposal example 5](https://github.com/cylc/cylc-admin/blob/master/docs/proposal-cylc-set.md#5-set-switch-tasks-at-an-optional-branch-point-to-direct-the-future-flow) > `$ cylc set //foo --out=a,succeeded --wait` > > The database will record: > > `foo: state=waiting,...

Undrafting this now as I think it implements the proposal fully. I've added functional tests that reproduce the examples in the proposal (IMO functional tests are needed for this sort...

(Almost there...)

(The only small hold-up here is: I removed the functional tests, and haven't finished replacing them with integration tests ... probably should have done that as a follow-up! - but...

> It might be worth inferring :succeeded to match graph behaviour? Yes, done.

Still not done adding tests. This afternoon I got stuck on the failing event-mail functional tests. They are failing for me locally too, but possibly for different reasons ... can...

> This illogical command had a curious outcome: > `cylc set --out=succeeded,failed` That's not illogical. A single running job can't both succeed and fail, but a task can fail and...

@oliver-sanders - I think I've address all your comments, and all the tests (except the usual MacOS ones) are passing now. I just need to add more integration tests tomorrow...

> 1. I don't think "set" is working with globs? It works with task name globs (`cylc set --out=succeeded "baz//1/*"`); I guess I hadn't tested with `//*`. [UPDATE]: actually I...

Although, nice work on the stress-testing, looks like I have my work cut out for me! Unfortunately I blew my time today looking at the globbing issues.