Unfortunately, the moment I start doing threaded or distributed computations it becomes markedly less helpful. With the example above, instead of smoothly updating in-step with the computation it has a tenancy to sit at 0% progress, and then jump to completion. This is particularly unhelpful when I have a multi-hour computation which I'd rather like to monitor.
I'm a big fan of
withprogress
, when it works. It's nice and easy to slot it into a pipeline like so:and I see a nice live-updating progress bar.
Unfortunately, the moment I start doing threaded or distributed computations it becomes markedly less helpful. With the example above, instead of smoothly updating in-step with the computation it has a tenancy to sit at 0% progress, and then jump to completion. This is particularly unhelpful when I have a multi-hour computation which I'd rather like to monitor.