Nick Crews

@derekeder Can you merge this tiny tiny fix?

Ehh, actually the solution of using the module where pytask.task() was called will fall apart if I have some make_task() wrapper in my_tasks.py, since then it will always be "my_tasks.py"....

yes that does work, but long term it would be nice if that wasn't required.

Can we take a step back and question: why does a task need to have a `module` attribute? For CLI-discovered tasks (eg a `def task_*` inside a `task_*.py` file, found...

Wait wait, if we can I really think we should try to avoid adding this argument to .task(). Why does a task need to have a module attribute? I'm wondering...

closing as stale, this doesn't seem to be ahuge priority...

This is true. Can you be more specific as to "extremely slow"? How long for how many files? Doing them all at once would require a larger re-structuring, it is...

that sounds like the correct thing to do. This is motivation to expose the "IF NOT EXISTS" functionality in .create_table(). Currently we have - overwrite=True: overwrite always - overwrite=False: write...

OK I think this is ready for review! I can break this into smaller commits if you want, I think I now have all the semantics figured out how I...

@kszucs I re-added some simple tests how does that look? I *can* add the tests for all the edge cases, but I would prefer not to. More of the actual...