New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
support dtype
kwarg for grouped reductions
#121
Comments
Is |
No, Here's a PR with useful tests and discussion: xarray-contrib/flox#131 |
I see. I think this is possible, though not trivial — currently we have a list of candidate signatures which numba will JIT compile for. We could generate new ones at runtime, but I'm not sure how numba handles a many-to-one mapping of inputs-to-outputs; so possibly we'd need to compile a new function... |
Ah I think this is lower priority. We can easily raise if dtype is provided and ask them to use another "engine" |
Yes or to convert first (albeit requires a copy) |
One question out of #128: To what extent do we want to convert types? Currently we always have the same input type as output type... ...except for But often that's not what we want — e.g. it is useful to have a float mean of ints, rather than an "int mean" (sum and then integer division). A few choices:
Unless folks disagree, I think we'll stay at the current state and see if we get traction and folks ask for a change... |
I think we can just ask the user to do so...
IMO this is OK.
Perhaps @shoyer has an opinion |
No description provided.
The text was updated successfully, but these errors were encountered: