[cond] error on closed over variables #99367
Closed
Conversation
This file contains 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
As reported in #90469, the implementation of inlining nested function branches for `cond` doesn't properly handle variables captured from outer scopes. This leads to some examples accidentally working, some others generating incorrect code that don't crash but do the wrong thing, and still others that outright crash because of references to non-existent variables. Properly supporting closed variables is tricky (see #91981 for an abandoned attempt). While this is definitely something we should be able to support longer term, for now it is better to explicitly error and suggest the fix to the user (amounting to rewriting branches to take closed variables explicitly). Differential Revision: [D45058621](https://our.internmc.facebook.com/intern/diff/D45058621/) [ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/99367
Note: Links to docs will display an error until the docs builds have been completed. ❌ 2 FailuresAs of commit 4a77cdc: NEW FAILURES - The following jobs have failed:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
This PR needs a labelIf your changes are user facing and intended to be a part of release notes, please use a label starting with If not, please add the To add a label, you can comment to pytorchbot, for example For more information, see |
As reported in #90469, the implementation of inlining nested function branches for `cond` doesn't properly handle variables captured from outer scopes. This leads to some examples accidentally working, some others generating incorrect code that don't crash but do the wrong thing, and still others that outright crash because of references to non-existent variables. Properly supporting closed variables is tricky (see #91981 for an abandoned attempt). While this is definitely something we should be able to support longer term, for now it is better to explicitly error and suggest the fix to the user (amounting to rewriting branches to take closed variables explicitly). Differential Revision: [D45058621](https://our.internmc.facebook.com/intern/diff/D45058621/) cc soumith voznesenskym penguinwu anijain2305 EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng Xia-Weiwen wenzhe-nrv jiayisunx desertfire [ghstack-poisoned]
avikchaudhuri
added a commit
that referenced
this pull request
Apr 17, 2023
Pull Request resolved: #99367 As reported in #90469, the implementation of inlining nested function branches for `cond` doesn't properly handle variables captured from outer scopes. This leads to some examples accidentally working, some others generating incorrect code that don't crash but do the wrong thing, and still others that outright crash because of references to non-existent variables. Properly supporting closed variables is tricky (see #91981 for an abandoned attempt). While this is definitely something we should be able to support longer term, for now it is better to explicitly error and suggest the fix to the user (amounting to rewriting branches to take closed variables explicitly). ghstack-source-id: 186352080 Differential Revision: [D45058621](https://our.internmc.facebook.com/intern/diff/D45058621/)
avikchaudhuri
requested review from
ezyang,
voznesenskym,
guangy10 and
tugsbayasgalan
guangy10
reviewed
Apr 18, 2023
ezyang
approved these changes
Apr 18, 2023
As reported in #90469, the implementation of inlining nested function branches for `cond` doesn't properly handle variables captured from outer scopes. This leads to some examples accidentally working, some others generating incorrect code that don't crash but do the wrong thing, and still others that outright crash because of references to non-existent variables. Properly supporting closed variables is tricky (see #91981 for an abandoned attempt). While this is definitely something we should be able to support longer term, for now it is better to explicitly error and suggest the fix to the user (amounting to rewriting branches to take closed variables explicitly). Differential Revision: [D45058621](https://our.internmc.facebook.com/intern/diff/D45058621/) cc soumith voznesenskym penguinwu anijain2305 EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng Xia-Weiwen wenzhe-nrv jiayisunx desertfire [ghstack-poisoned]
avikchaudhuri
added a commit
that referenced
this pull request
Apr 18, 2023
Pull Request resolved: #99367 As reported in #90469, the implementation of inlining nested function branches for `cond` doesn't properly handle variables captured from outer scopes. This leads to some examples accidentally working, some others generating incorrect code that don't crash but do the wrong thing, and still others that outright crash because of references to non-existent variables. Properly supporting closed variables is tricky (see #91981 for an abandoned attempt). While this is definitely something we should be able to support longer term, for now it is better to explicitly error and suggest the fix to the user (amounting to rewriting branches to take closed variables explicitly). ghstack-source-id: 186391673 Differential Revision: [D45058621](https://our.internmc.facebook.com/intern/diff/D45058621/)
As reported in #90469, the implementation of inlining nested function branches for `cond` doesn't properly handle variables captured from outer scopes. This leads to some examples accidentally working, some others generating incorrect code that don't crash but do the wrong thing, and still others that outright crash because of references to non-existent variables. Properly supporting closed variables is tricky (see #91981 for an abandoned attempt). While this is definitely something we should be able to support longer term, for now it is better to explicitly error and suggest the fix to the user (amounting to rewriting branches to take closed variables explicitly). Differential Revision: [D45058621](https://our.internmc.facebook.com/intern/diff/D45058621/) cc soumith voznesenskym penguinwu anijain2305 EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng Xia-Weiwen wenzhe-nrv jiayisunx desertfire [ghstack-poisoned]
avikchaudhuri
added a commit
that referenced
this pull request
Apr 18, 2023
Pull Request resolved: #99367 As reported in #90469, the implementation of inlining nested function branches for `cond` doesn't properly handle variables captured from outer scopes. This leads to some examples accidentally working, some others generating incorrect code that don't crash but do the wrong thing, and still others that outright crash because of references to non-existent variables. Properly supporting closed variables is tricky (see #91981 for an abandoned attempt). While this is definitely something we should be able to support longer term, for now it is better to explicitly error and suggest the fix to the user (amounting to rewriting branches to take closed variables explicitly). ghstack-source-id: 186457954 Differential Revision: [D45058621](https://our.internmc.facebook.com/intern/diff/D45058621/)
tugsbayasgalan
approved these changes
Apr 18, 2023
As reported in #90469, the implementation of inlining nested function branches for `cond` doesn't properly handle variables captured from outer scopes. This leads to some examples accidentally working, some others generating incorrect code that don't crash but do the wrong thing, and still others that outright crash because of references to non-existent variables. Properly supporting closed variables is tricky (see #91981 for an abandoned attempt). While this is definitely something we should be able to support longer term, for now it is better to explicitly error and suggest the fix to the user (amounting to rewriting branches to take closed variables explicitly). Differential Revision: [D45058621](https://our.internmc.facebook.com/intern/diff/D45058621/) cc soumith voznesenskym penguinwu anijain2305 EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng Xia-Weiwen wenzhe-nrv jiayisunx desertfire [ghstack-poisoned]
avikchaudhuri
added a commit
that referenced
this pull request
Apr 19, 2023
Pull Request resolved: #99367 As reported in #90469, the implementation of inlining nested function branches for `cond` doesn't properly handle variables captured from outer scopes. This leads to some examples accidentally working, some others generating incorrect code that don't crash but do the wrong thing, and still others that outright crash because of references to non-existent variables. Properly supporting closed variables is tricky (see #91981 for an abandoned attempt). While this is definitely something we should be able to support longer term, for now it is better to explicitly error and suggest the fix to the user (amounting to rewriting branches to take closed variables explicitly). ghstack-source-id: 186556432 Differential Revision: [D45058621](https://our.internmc.facebook.com/intern/diff/D45058621/)
|
Issue for follow up: #99401 |
|
@pytorchbot merge -f "unrelated ci failures (doctr)" |
Merge startedYour change will be merged immediately since you used the force (-f) flag, bypassing any CI checks (ETA: 1-5 minutes). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
avikchaudhuri
added a commit
that referenced
this pull request
Apr 25, 2023
As of #99367 we error when cond branches look up closed vars. The suggested fix is to add these closed vars as args to the branches. However, while this works for tensor vars (and also primitive vars by explicit wrapping), this is impossible to do for function vars. Moreover, function vars are OK because we trace through them. So relaxing this restriction for function vars is a strict win. Differential Revision: [D45287893](https://our.internmc.facebook.com/intern/diff/D45287893/) [ghstack-poisoned]
avikchaudhuri
added a commit
that referenced
this pull request
Apr 25, 2023
As of #99367 we error when cond branches look up closed vars. The suggested fix is to add these closed vars as args to the branches. However, while this works for tensor vars (and also primitive vars by explicit wrapping), this is impossible to do for function vars. Moreover, function vars are OK because we trace through them. So relaxing this restriction for function vars is a strict win. Differential Revision: [D45287893](https://our.internmc.facebook.com/intern/diff/D45287893/) ghstack-source-id: 187112368 Pull Request resolved: #100013
avikchaudhuri
added a commit
that referenced
this pull request
Apr 25, 2023
As of #99367 we error when cond branches look up closed vars. The suggested fix is to add these closed vars as args to the branches. However, while this works for tensor vars (and also primitive vars by explicit wrapping), this is impossible to do for function vars. Moreover, function vars are OK because we trace through them. So relaxing this restriction for function vars is a strict win. Differential Revision: [D45287893](https://our.internmc.facebook.com/intern/diff/D45287893/) cc soumith voznesenskym penguinwu anijain2305 EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng Xia-Weiwen wenzhe-nrv jiayisunx desertfire [ghstack-poisoned]
avikchaudhuri
added a commit
that referenced
this pull request
Apr 25, 2023
Pull Request resolved: #100013 As of #99367 we error when cond branches look up closed vars. The suggested fix is to add these closed vars as args to the branches. However, while this works for tensor vars (and also primitive vars by explicit wrapping), this is impossible to do for function vars. Moreover, function vars are OK because we trace through them. So relaxing this restriction for function vars is a strict win. ghstack-source-id: 187142449 Differential Revision: [D45287893](https://our.internmc.facebook.com/intern/diff/D45287893/)
pytorchmergebot
pushed a commit
that referenced
this pull request
Apr 26, 2023
As of #99367 we error when cond branches look up closed vars. The suggested fix is to add these closed vars as args to the branches. However, while this works for tensor vars (and also primitive vars by explicit wrapping), this is impossible to do for function vars. Moreover, function vars are OK because we trace through them. So relaxing this restriction for function vars is a strict win. Differential Revision: [D45287893](https://our.internmc.facebook.com/intern/diff/D45287893/) Pull Request resolved: #100013 Approved by: https://github.com/tugsbayasgalan
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Stack from ghstack (oldest at bottom):
As reported in #90469, the implementation of inlining nested function branches for
conddoesn't properly handle variables captured from outer scopes. This leads to some examples accidentally working, some others generating incorrect code that don't crash but do the wrong thing, and still others that outright crash because of references to non-existent variables.Properly supporting closed variables is tricky (see #91981 for an abandoned attempt). While this is definitely something we should be able to support longer term, for now it is better to explicitly error and suggest the fix to the user (amounting to rewriting branches to take closed variables explicitly).
Differential Revision: D45058621
cc @soumith @voznesenskym @penguinwu @anijain2305 @EikanWang @jgong5 @Guobing-Chen @XiaobingSuper @zhuhaozhe @blzheng @Xia-Weiwen @wenzhe-nrv @jiayisunx @desertfire