Add a private API that lists overridable functions #33182
Assignees
Labels
enhancement
Not as big of a feature, but technically not a bug. Should be easy to fix
module: tests
Issues related to tests (not the torch.testing module)
triaged
This issue has been looked at a team member, and triaged and prioritized into an appropriate module
Comments
|
On the list of functions in the documentation: I think it's sufficient, if it'll be kept up to date. A private API that's also used in testing could be a good forcing function to make sure that list is always complete. |
ngoldbaum
changed the title
gchanan
added
module: tests
gchanan
removed
the
needs research
|
we think this is reasonable and would accept a PR that implements it. |
|
I'm going to do this but I'll wait for #32799 to land before putting in a PR that fixes this. |
ttumiel
pushed a commit
to ttumiel/pytorch
that referenced
this issue
Mar 4, 2020
…rch#33791) Summary: Fixes pytorch#33182 This adds private API functions that developers of types that implement `__torch_function__` can use to ensure full coverage of the subset of the PyTorch API that can be overrided. I've refactored some of the code in the tests into a new `torch._overrides.get_overridable_functions` function. I've also changed `TENSOR_LIKE_TORCH_OVERRIDES` into `torch._overrides.get_testing_overrides` and `IGNORED_TORCH_FUNCTIONS` into `torch._overrides.get_ignored_functions`. Making these two static global variables in the tests into functions should allow rewriting their implementation to construct their return values instead of just statically defining the return value as is done here. Currently that is blocked on not being able to inspect function signatures of compiled kernels in PyTorch (see pytorch#28233). See the docs I've added for usage examples of these new functions. I also refactored the existing override tests to make use of these new functions, which should be a good forcing function to make sure they're kept up-to-date. Finally, while working on this I discovered that `TestTorchFunctionOverrides.test_mean` and `TestTorchFunctionOverrides.test_mm` weren't ever being run because they were getting clobbered by the other dynamically generated override tests. I fixed that by renaming the tests and then fixing the actual test code. I've verified that all the subclassing semantics is correct and that the updated test answers are correct. I'm happy to put the fixes to the existing tests in as a separate pull request if that would be easier to review. ping cpuhrsch since the feature request originally came from them. Pull Request resolved: pytorch#33791 Differential Revision: D20195053 Pulled By: cpuhrsch fbshipit-source-id: 1585f4e405f5223932b410eae03a288dc8eb627e
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
🚀 Feature
Add a way for library authors who depend on pytorch to get a list of all functions that they would need to implement to fully wrap the API surface that is overridable via
__torch_function__.Motivation
See e.g. #27064 (comment).
Pitch
Currently it's not straightforward for authors of types that implement
__torch_function__to test that they fully cover the API that is overridable via__torch_function__. It's important that they are able to wrap as much of the API as possible because once you have written a type that implements__torch_function__users of that type will see aTypeErrorif the__torch_function__implementation returnsNotImplementedfor unwrapped functions.It would be nice if there were some private API that gave developers access to the functions that are overridable. It might also be nice to include the function signatures in an easily introspectable format as that makes it easier to write tests for the overrides.
Alternatives
It might also be sufficient to just have a list of functions in the documentation.
The text was updated successfully, but these errors were encountered: