Improve error when torchvision C++ ops fail to load - #9649
Open
YeonwooSung wants to merge 1 commit into
Open
Conversation
Skip register_fake for nms/qnms when compiled ops are missing so importing an unbuilt source tree no longer fails with "operator torchvision::nms does not exist". Cache the extension load exception and include it in _assert_has_ops(). Fixes pytorch#9337
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/vision/9649
Note: Links to docs will display an error until the docs builds have been completed. This comment was automatically generated by Dr. CI and updates every 15 minutes. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Importing an unbuilt torchvision source tree (or a broken / ABI-mismatched wheel) currently fails at import time with:
This happens because
_meta_registrations.pyregisters fake kernels fortorchvision::nms/torchvision::qnmsunconditionally.torch.library.register_fakerequires the C++ operator to already exist, so the failure happens beforenms()can reach_assert_has_ops()._load_library()also swallowedImportError/OSErrorand returnedFalse, so the original missing.so/ wheel-mismatch reason was lost.Changes
register_fakefornms/qnmswithtorchvision.extension._has_ops(), matchingregister_meta. Importing torchvision without compiled ops now succeeds; callingnms()then hits_assert_has_ops()._EXTENSION_LOAD_ERROR(preferring the_C_stableerror so a later optional-extension failure does not hide it) and include it in_assert_has_ops().test/test_extension.pycovering the cached load error,_assert_has_ops()message, and theregister_fakeguard.Fixes #9337