thoughts on checkpoints

[ctb]

my usual advice: if you get trapped by checkpoints, start by writing multiple different snakefiles and running them in serial. this is not as efficient as using checkpoints, necessarily, but might get you there.

I wonder if it's possible to do an illustrative example where we write a snakefile starting with using checkpoints from the very first rule? does that even make sense?

[ctb]

hmm, I wonder if I could put together some generic tracing and debugging stuff that would work for understanding and debugging checkpoints?