Build Python module and tests CPU mode with Bazel in CI#78221
Build Python module and tests CPU mode with Bazel in CI#78221dreiss wants to merge 7 commits intogh/dreiss/119/basefrom
Conversation
This can potentially catch issues like #77002 before they hit Meta-internal CI, because Bazel is more strict about header visibility than (our current setup of) CMake. [ghstack-poisoned]
🔗 Helpful links
✅ No Failures (0 Pending)As of commit 09ff91d (more details on the Dr. CI page): Expand to see more💚 💚 Looks good so far! There are no failures yet. 💚 💚 This comment was automatically generated by Dr. CI (expand for details).Please report bugs/suggestions to the (internal) Dr. CI Users group. |
|
CI should fail initially because #77002 is currently in viable/strict, which I'm based on. |
This can potentially catch issues like #77002 before they hit Meta-internal CI, because Bazel is more strict about header visibility than (our current setup of) CMake. [ghstack-poisoned]
This can potentially catch issues like #77002 before they hit Meta-internal CI, because Bazel is more strict about header visibility than (our current setup of) CMake. [ghstack-poisoned]
This can potentially catch issues like #77002 before they hit Meta-internal CI, because Bazel is more strict about header visibility than (our current setup of) CMake. [ghstack-poisoned]
This can potentially catch issues like #77002 before they hit Meta-internal CI, because Bazel is more strict about header visibility than (our current setup of) CMake. [ghstack-poisoned]
This can potentially catch issues like #77002 before they hit Meta-internal CI, because Bazel is more strict about header visibility than (our current setup of) CMake. [ghstack-poisoned]
|
Okay, I finally got this version to fail with the proper error ("torch/csrc/Module.cpp:10:10: fatal error: ATen/cuda/CUDAConfig.h: No such file or directory"). https://github.com/pytorch/pytorch/runs/6596140899?check_suite_focus=true Now I'm going to rebase, make sure it passes, and see what the perf impact is. |
|
It looks like the extra command took ~5:30, which is more than I expected, but still not a problem. I'm going to combine the two CPU builds to hopefully unlock a bit more parallelism, then land. |
This can potentially catch issues like #77002 before they hit Meta-internal CI, because Bazel is more strict about header visibility than (our current setup of) CMake. [ghstack-poisoned]
|
@pytorchbot help |
|
To see all options for pytorchbot, please refer to this page. |
|
@pytorchbot merge -g please |
|
Hey @dreiss. |
Summary: This can potentially catch issues like #77002 before they hit Meta-internal CI, because Bazel is more strict about header visibility than (our current setup of) CMake. Pull Request resolved: #78221 Approved by: https://github.com/malfet Test Plan: contbuild & OSS CI, see https://hud.pytorch.org/commit/pytorch/pytorch/8449ac770c29a52257a3f2842c1dfd06d2c2a497 Reviewed By: mehtanirav Differential Revision: D36707271 Pulled By: dreiss fbshipit-source-id: d9c3d17e9227b7660a0819e9894767035d10863f
Stack from ghstack:
This can potentially catch issues like #77002 before they hit
Meta-internal CI, because Bazel is more strict about header visibility
than (our current setup of) CMake.