[Comments PTS 13] Rethinking Python decorators

[joaomcteixeira]

Original post: https://pythonicthoughtssnippets.github.io/2020/08/09/PTS13-rethinking-python-decorators.html
Post on DEV: https://dev.to/joaomcteixeira/rethinking-python-decorators-2g53

[caseyclements]

Thank you for this post! It was precisely what I needed. I was able to refactor an existing library containing a combination of nested lambdas, decorators, and inner functions using plain pickle, and then distribute using dask.

[joaomcteixeira]

Hi @caseyclements !

Thanks very much for your comment. I am really happy you could use that strategy to refact on your side. Not much people talk about these intricate topics and there is definitively a leap from using python to write scripts and using python to write whole and structured programs.

Have you encountered any difficulties or additions beyond what I refer in my post? Glad to discuss further 👍

Cheers,

[bernstei]

Thanks for the interesting explanation. Using your functools.partial idea I was able to get a convenient pickleable syntax for a decorator with arguments. However, can you think of any way to get the docstring for the resulting function to be informative, rather than just the meaningless docstring for partial that results?

[joaomcteixeira]

Hi @bernstei, thanks so much for your feedback. It really means a lot to me you enjoyed the post and that it was useful for you. Replacing the partialized function with the original docstring would be "a lie" because some parameters are already input. However, you can actually rewrite the partialized function docstring easily:

pfunc = functools.partial(myfunc, arg1=10)
pfunc.__doc__ = """Does what myfunc does but with arg1 = 10""" # you can go multiline here.
help(pfunc)  # shows you the new docstring.

Cheers,

[mutricyl]

Thanks indeed for your work. I needed a multiprocessing safe "decorator". The partial strategy works great !

[joaomcteixeira]

Many thanks @mutricyl for your feedback. All the best!