Iter performer #81
Iter performer #81
Conversation
Tomatosoup97
force-pushed
the
iter-performer
branch
from
667e43d to
5d8e4cc
Compare
|
Thanks for the contribution! I've started looking at it, but it's going to take me a while to absorb it. |
There was a problem hiding this comment.
@Tomatosoup97 Can you describe a little more in-depth what your use-case for this functionality is? I can basically understand what the code is doing, but not sure where it would be useful. That would help produce better docs for it, too.
One of the things that strikes me is that it stores state in the performer functions. The way that you'd have to set up the performers seems at-odds with the general pattern I see/prefer for writing Effect-based code -- basically, that performers are all set up in one central place, near the main function of your application (along with the call to sync_perform). If I understand this correctly, code that uses this would want the state to have a finer-grained scope than a global one.
I think a way to get around this would be to introduce a third intent: one to create some state, along with the others to append to it and read from it. The result of performing the "state creation" intent would return a fresh list, then the other intents would operate on that list. This new intent could be abstracted away from any user of the decorator that you wrote, which could take care of that for them and (I think) maintain an identical API to the current one.
Still, though, I have no idea what this would actually be useful for, so I'm not sure if this suggestion really makes sense.
| """Provides retrieval of gathered values""" | ||
| return xs | ||
|
|
||
| return partial(performer, xs), partial(retriever, xs) |
There was a problem hiding this comment.
since xs is in the closure anyway, you don't need to explicitly take it as a parameter to performer and retriever, and don't need to explicitly partial it to bind it to those functions.
|
Closing this for now. If you ever feel like reopening the discussion, just comment here and I can reopen the PR. |
Hello. I've found myself in need to have actual generators among the effectful code - or in other words, to have a standard python generator that uses effects. This library already relies on
yieldsyntax, so it's not straighforward, and thus I wrote a peformer to do that.I think that the code turned out to be generic enough to (maybe) be incorporated into the upstream - the need for actual generators seems like a simple and reocurring thing.
Obvious issues with the PR:
testtoolsfor testing)toolz- I don't really need it, since I only usecurryfrom there. This could be rewritten topartials to cut the dependency.I did not bother myself yet to address these issues. Instead, at first I would like to hear your opinion and whether you would like to see something like that in the repo, and then I can polish the details. Tests are here natural usage examples.