Bbonev/wheels #127
Open
Bbonev/wheels #127
Conversation
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
…ble when setup.py is run
- Remove redundant 'Prepare build environment' step from GitHub Actions - Move PyTorch installation to CIBW_BEFORE_BUILD (runs inside manylinux container) - Remove redundant before-build from pyproject.toml - This eliminates double environment setup and makes builds more efficient
- Add --exclude flags for libc10.so, libtorch.so, libtorch_cpu.so, libtorch_cuda.so - This prevents auditwheel from trying to bundle PyTorch libraries - PyTorch should be installed separately by users, not bundled in wheels - Fixes 'Cannot repair wheel, because required library libc10.so could not be located' error
- Add --exclude flags for numpy-related libraries (libopenblas.so, liblapack.so, libgfortran.so, etc.) - Add --exclude flags for system libraries (libm.so, libc.so, libgomp.so) - These libraries should be provided by the system or installed separately by users - Prevents bloated wheels and potential conflicts with system libraries - Follows best practices for Python wheel packaging
- Remove --exclude for system libraries (libm.so, libc.so, libgomp.so) - Keep --exclude for PyTorch libraries (libc10.so, libtorch.so, etc.) - Keep --exclude for numpy libraries (libopenblas.so, liblapack.so, etc.) - System libraries are part of manylinux standard and should be bundled - PyTorch/numpy libraries should be installed separately by users - This ensures wheels work on systems with different library versions
- Add libtorch_python.so (fixes current error) - Add libtorch_global_deps.so, libtorch_cuda_cpp.so, libtorch_cuda_cu.so - Keep existing exclusions for core PyTorch and numpy libraries - This should resolve the 'libtorch_python.so could not be located' error - Test with smaller list before expanding to comprehensive exclusion
…iants - Keep torch>=2.4.0 in build-system.requires (required for setup.py) - Add --force-reinstall to PyTorch installation in CIBW_BEFORE_BUILD - This ensures the specific PyTorch version with correct CUDA variant is installed - Overrides the generic torch>=2.4.0 installed by build system - Fixes build system pip install failure
- Remove --force-reinstall flags from PyTorch installation - PyTorch is installed FIRST in CIBW_BEFORE_BUILD (before build system runs) - Build system will find PyTorch already installed (correct version) - This ensures tight control over PyTorch version without conflicts - Cleaner approach: install PyTorch first, then let build system use it
- Fix trailing comma in dependencies array (line 40) - Add missing torch>=2.4.0 dependency - Resolves tomllib.TOMLDecodeError: Invalid value
- Add CIBW_BUILD_FRONTEND_ARGS: --no-build-isolation - Prevents build system from creating separate isolated environment - Build system will use PyTorch installed by CIBW_BEFORE_BUILD - Fixes 'pip install' failure in isolated build environment - Ensures build system uses correct PyTorch version with CUDA variant
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.
No description provided.