Replace calls to python setup.py sdist with `python -m build --sdis…

…t` (#7069)

With future changes coming to pip/python/etc, we need to modify to no
longer call `python setup.py ...` and replace it instead:
https://packaging.python.org/en/latest/guides/modernize-setup-py-project/#should-setup-py-be-deleted


![image](https://github.com/user-attachments/assets/ea39ef7b-3cbe-4916-86f0-bc46a5fce96d)

This means we need to install the build package which is added here as
well.

Additionally, we pass the `--sdist` flag to only build the sdist rather
than the wheel as well here.

---------

Signed-off-by: Logan Adams <[email protected]>

.github/workflows/no-torch.yml

@@ -32,11 +32,12 @@ jobs:
         run: |
           pip uninstall torch --yes
           pip install setuptools
+          pip install build
           pip list
 
       - name: Build deepspeed
         run: |
-          DS_BUILD_STRING=" " python setup.py sdist
+          DS_BUILD_STRING=" " python -m build --sdist
 
       - name: Open GitHub issue if nightly CI fails
         if: ${{ failure() && (github.event_name == 'schedule') }}

.github/workflows/release.yml

@@ -26,7 +26,8 @@ jobs:
     - name: Build DeepSpeed
       run: |
         pip install setuptools
-        DS_BUILD_STRING=" " python setup.py sdist
+        pip install build
+        DS_BUILD_STRING=" " python -m build --sdist
     - name: Publish to PyPI
       uses: pypa/gh-action-pypi-publish@release/v1
       with:

build_win.bat

@@ -11,6 +11,6 @@ set DS_BUILD_GDS=0
 set DS_BUILD_RAGGED_DEVICE_OPS=0
 set DS_BUILD_SPARSE_ATTN=0
 
-python setup.py bdist_wheel
+python -m build --wheel --no-isolation
 
 :end

docs/_tutorials/advanced-install.md

@@ -84,7 +84,7 @@ This should complete the full build 2-3 times faster. You can adjust `-j` to spe
 You can also build a binary wheel and install it on multiple machines that have the same type of GPUs and the same software environment (CUDA toolkit, PyTorch, Python, etc.)
 
 ```bash
-DS_BUILD_OPS=1 python setup.py build_ext -j8 bdist_wheel
+DS_BUILD_OPS=1 python -m build --wheel --no-isolation --config-setting="--build-option=build_ext" --config-setting="--build-option=-j8"
 ```
 
 This will create a pypi binary wheel under `dist`, e.g., ``dist/deepspeed-0.3.13+8cd046f-cp38-cp38-linux_x86_64.whl`` and then you can install it directly on multiple machines, in our example:

docs/_tutorials/ds-sequence.md

@@ -111,7 +111,7 @@ pip install .
 cd ${WORK_DIR}
 git clone -b v1.0.4 https://github.com/HazyResearch/flash-attention
 cd flash-attention
-python setup.py install
+python -m pip install .
 ```
 
 You may also want to ensure your model configuration is compliant with FlashAttention's requirements. For instance, to achieve optimal performance, the head size should be divisible by 8. Refer to the FlashAttention documentation for more details.

install.sh

@@ -152,7 +152,7 @@ if [ ! -f $hostfile ]; then
 fi
 
 echo "Building deepspeed wheel"
-python setup.py $VERBOSE bdist_wheel
+python -m build $VERBOSE --wheel --no-isolation
 
 if [ "$local_only" == "1" ]; then
     echo "Installing deepspeed"

release/release.sh

@@ -38,7 +38,7 @@ if [ $? != 0 ]; then
     exit 1
 fi
 
-DS_BUILD_STRING="" python setup.py sdist
+DS_BUILD_STRING="" python -m build --sdist
 
 if [ ! -f dist/deepspeed-${version}.tar.gz ]; then
     echo "prepared version does not match version given ($version), bump version first?"

setup.py

@@ -233,7 +233,7 @@ def op_enabled(op_name):
 version_str = open('version.txt', 'r').read().strip()
 
 # Build specifiers like .devX can be added at install time. Otherwise, add the git hash.
-# Example: DS_BUILD_STRING=".dev20201022" python setup.py sdist bdist_wheel.
+# Example: `DS_BUILD_STRING=".dev20201022" python -m build --no-isolation`.
 
 # Building wheel for distribution, update version file.
 if is_env_set('DS_BUILD_STRING'):