diff options
| author | anderspapitto <anderspapitto@gmail.com> | 2018-06-12 14:57:29 -0700 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2018-06-12 14:57:29 -0700 |
| commit | fcd9af8a257f83450ba82cccd227fe43bde7c879 (patch) | |
| tree | 49968013bdc5aa8c44e6989cf9d26f4c5474321e /torch/_thnn | |
| parent | ffffee6aa99331fb6bed453009350805615499a0 (diff) | |
| download | pytorch-fcd9af8a257f83450ba82cccd227fe43bde7c879.tar.gz pytorch-fcd9af8a257f83450ba82cccd227fe43bde7c879.tar.bz2 pytorch-fcd9af8a257f83450ba82cccd227fe43bde7c879.zip | |
changes to support ATen code generation inside fbcode (#8397)
* Back out "Back out "Add support for generating ATen files during fbcode build""
Original commit changeset: 7b8de22d1613
I'm re-sending this diff exactly as it was approved and
committed. Fixes to support @mode/opt will be sent separately for ease
of review.
* Enable building //caffe2:torch with @mode/opt
In @mode/opt, python runs out of a PAR, which breaks a lot of
assumptions in the code about where templates/ folders live relative
to __file__. Rather than introduce hacks with parutil, I simply turn
template_path into a parameter for all the relevant functions and
thread it through from the top level.
Diffstat (limited to 'torch/_thnn')
| -rw-r--r-- | torch/_thnn/utils.py | 8 |
1 files changed, 6 insertions, 2 deletions
diff --git a/torch/_thnn/utils.py b/torch/_thnn/utils.py index ab164e02e6..84c93a8180 100644 --- a/torch/_thnn/utils.py +++ b/torch/_thnn/utils.py @@ -2,8 +2,12 @@ import os import itertools import importlib -THNN_H_PATH = os.path.join(os.path.dirname(__file__), '..', 'lib', 'THNN.h') -THCUNN_H_PATH = os.path.join(os.path.dirname(__file__), '..', 'lib', 'THCUNN.h') +# in fbcode, this fails in some cases, but we don't need it, therefore the try-catch +try: + THNN_H_PATH = os.path.join(os.path.dirname(__file__), '..', 'lib', 'THNN.h') + THCUNN_H_PATH = os.path.join(os.path.dirname(__file__), '..', 'lib', 'THCUNN.h') +except Exception: + pass def _unpickle_backend(backend_name): |