I had an idea for a feature when I was debugging an old playbook on a new type of host.
I would love to be able to do per-action confirmation before execution for a playbook.
something like:
ansible-playbook --ask myplaybook.yml
and then have it ask for a 'y/n' before every action it will take. So I can step around a known problem, for example.
Is that completely ridiculous or something folks might find useful? Seems like it would only require a pre-execution callback of some kind - which I'm not sure we have yet - but might be do-able.
Using 'name' avoids adding new syntax and having to explain to users what the difference between a 'name' and a 'label' is. Using 'name' also means '--start-at-task' can be used on an unmodified playbook. Plus I'd generally think task names are likely to be unique across a playbook run (or at least that would be a best practice). Can think of a case where it makes much sense to use the same name twice -- it would make interpreting the output a bit tricky.
I'd go for using fnmatch on name (or action if name is absent) if people want to step through (and using breakpoints) they can add something unique to the name if they really want to have a guaranteed single breakpoint.
The clear benefit is that if you want to e.g. debug an NTP related issue, there's a high chance that using 'ntp' in the fnmatch will simply do the right thing out of the box.
In this case I would even go for a case-insensitive match.