[Release] Bump version to v0.1.6.post2 (#1160)

* [Release] Update README and VERSION for v0.1.6.post2 compatibility with Python 3.8

* [Enhancement] Update packaging configuration and Docker scripts for multi-architecture support

* Add allowlist for TVM, CUTLASS, and Composable Kernel items in pyproject.toml
* Enhance docker_local_distribute.sh to support cross-architecture builds using docker buildx
* Modify pypi.manylinux.Dockerfile to accept TARGETARCH argument for better architecture handling

* [Enhancement] Improve Docker scripts and build process for multi-architecture support

* Update .gitignore to include dist directories
* Refactor docker_local_distribute.sh for better cross-architecture handling and error management
* Enhance docker_pypi_distribute.sh to support multi-architecture builds with docker buildx
* Modify pypi_distribution.sh to clean up additional directories
* Update pypi.manylinux.Dockerfile for improved environment configuration and architecture handling

* fix

* Remove outdated classifier for Artificial Intelligence from pyproject.toml

* Update pyproject.toml classifiers and modify Docker distribution scripts for clarity

* Add new classifier for Artificial Intelligence in pyproject.toml
* Rename output directories in docker_local_distribute.sh and docker_pypi_distribute.sh for better context

.gitignore

@@ -20,6 +20,7 @@
 debug/
 build/
 *dist/
+dist*/
 wheelhouse/
 __pycache__
 nnfusion.tar.gz

README.md

@@ -13,6 +13,7 @@ Tile Language (**tile-lang**) is a concise domain-specific language designed to
 <img src=./images/MatmulExample.png />
 
 ## Latest News
+- 10/30/2025 📦: We have released v0.1.6.post2, which is the last version compatible with Python 3.8.
 - 10/07/2025 🍎: Added Apple Metal Device support, check out [Pull Request #799](https://github.com/tile-ai/tilelang/pull/799) for details.
 - 09/29/2025  🎉: Thrilled to announce that ​​AscendC​​ and ​Ascend​NPU IR​​ backends targeting Huawei Ascend chips are now supported!
 Check out the preview here:

VERSION

-0.1.6.post1
+0.1.6.post2

maint/scripts/docker_local_distribute.sh

-set -eux
+#!/usr/bin/env bash
+set -euxo pipefail
 
-# Get the CUDA version from the command line
 IMAGE="tilelang-builder:manylinux"
-docker build . -f "$(dirname "${BASH_SOURCE[0]}")/pypi.manylinux.Dockerfile" --tag ${IMAGE}
 
-script="sh maint/scripts/local_distribution.sh"
+HOST_UNAME=$(uname -m)
+case "$HOST_UNAME" in
+  x86_64) TARGETARCH=amd64 ;;
+  aarch64|arm64) TARGETARCH=arm64 ;;
+  *) echo "Unsupported architecture: $HOST_UNAME" >&2; exit 1 ;;
+esac
 
-docker run --rm -v $(pwd):/tilelang ${IMAGE} /bin/bash -c "$script"
+if docker buildx version >/dev/null 2>&1; then
+  if docker info >/dev/null 2>&1; then
+    docker run --rm --privileged tonistiigi/binfmt --install amd64,arm64 >/dev/null 2>&1 || true
+  fi
+
+  if ! docker buildx inspect multi >/dev/null 2>&1; then
+    docker buildx create --name multi --driver docker-container --use >/dev/null 2>&1 || true
+  else
+    docker buildx use multi >/dev/null 2>&1 || true
+  fi
+  docker buildx inspect --bootstrap >/dev/null 2>&1 || true
+
+  for ARCH in amd64 arm64; do
+    TAG_PLATFORM="linux/${ARCH}"
+    TAG_IMAGE="${IMAGE}-${ARCH}"
+
+    docker buildx build \
+      --platform "${TAG_PLATFORM}" \
+      --build-arg TARGETARCH="${ARCH}" \
+      -f "$(dirname "${BASH_SOURCE[0]}")/pypi.manylinux.Dockerfile" \
+      -t "${TAG_IMAGE}" \
+      --load \
+      .
+
+    script="sh maint/scripts/local_distribution.sh"
+    docker run --rm \
+      --platform "${TAG_PLATFORM}" \
+      -v "$(pwd):/tilelang" \
+      "${TAG_IMAGE}" \
+      /bin/bash -lc "$script"
+
+    if [ -d dist ]; then
+      mv -f dist "dist-local-${ARCH}"
+    fi
+  done
+
+else
+  echo "docker buildx not found; building only host arch: ${TARGETARCH}" >&2
+  TAG_IMAGE="${IMAGE}-${TARGETARCH}"
+  TAG_PLATFORM="linux/${TARGETARCH}"
+
+  docker build \
+    --build-arg TARGETARCH="$TARGETARCH" \
+    -f "$(dirname "${BASH_SOURCE[0]}")/pypi.manylinux.Dockerfile" \
+    -t "${TAG_IMAGE}" \
+    .
+
+  script="sh maint/scripts/local_distribution.sh"
+  docker run --rm \
+    --platform "${TAG_PLATFORM}" \
+    -v "$(pwd):/tilelang" \
+    "${TAG_IMAGE}" \
+    /bin/bash -lc "$script"
+
+  if [ -d dist ]; then
+    mv -f dist "dist-local-${TARGETARCH}"
+  fi
+fi

maint/scripts/docker_pypi_distribute.sh

-set -eux
+#!/usr/bin/env bash
+set -euxo pipefail
 
-# Get the CUDA version from the command line
 IMAGE="tilelang-builder:manylinux"
-docker build . -f "$(dirname "${BASH_SOURCE[0]}")/pypi.manylinux.Dockerfile" --tag ${IMAGE}
 
-script="sh maint/scripts/pypi_distribution.sh"
+HOST_UNAME=$(uname -m)
+case "$HOST_UNAME" in
+  x86_64) TARGETARCH=amd64 ;;
+  aarch64|arm64) TARGETARCH=arm64 ;;
+  *) echo "Unsupported architecture: $HOST_UNAME" >&2; exit 1 ;;
+esac
 
-docker run --rm -v $(pwd):/tilelang -w /tilelang ${IMAGE} /bin/bash -c "$script"
+if docker buildx version >/dev/null 2>&1; then
+  if docker info >/dev/null 2>&1; then
+    docker run --rm --privileged tonistiigi/binfmt --install amd64,arm64 >/dev/null 2>&1 || true
+  fi
+
+  if ! docker buildx inspect multi >/dev/null 2>&1; then
+    docker buildx create --name multi --driver docker-container --use >/dev/null 2>&1 || true
+  else
+    docker buildx use multi >/dev/null 2>&1 || true
+  fi
+  docker buildx inspect --bootstrap >/dev/null 2>&1 || true
+
+  for ARCH in amd64 arm64; do
+    TAG_PLATFORM="linux/${ARCH}"
+    TAG_IMAGE="${IMAGE}-${ARCH}"
+
+    docker buildx build \
+      --platform "${TAG_PLATFORM}" \
+      --build-arg TARGETARCH="${ARCH}" \
+      -f "$(dirname "${BASH_SOURCE[0]}")/pypi.manylinux.Dockerfile" \
+      -t "${TAG_IMAGE}" \
+      --load \
+      .
+
+    script="sh maint/scripts/pypi_distribution.sh"
+    docker run --rm \
+      --platform "${TAG_PLATFORM}" \
+      -v "$(pwd):/tilelang" \
+      "${TAG_IMAGE}" \
+      /bin/bash -lc "$script"
+
+    if [ -d dist ]; then
+      mv -f dist "dist-pypi-${ARCH}"
+    fi
+  done
+
+else
+  echo "docker buildx not found; building only host arch: ${TARGETARCH}" >&2
+  TAG_IMAGE="${IMAGE}-${TARGETARCH}"
+  TAG_PLATFORM="linux/${TARGETARCH}"
+
+  docker build \
+    --build-arg TARGETARCH="$TARGETARCH" \
+    -f "$(dirname "${BASH_SOURCE[0]}")/pypi.manylinux.Dockerfile" \
+    -t "${TAG_IMAGE}" \
+    .
+
+  script="sh maint/scripts/pypi_distribution.sh"
+  docker run --rm \
+    --platform "${TAG_PLATFORM}" \
+    -v "$(pwd):/tilelang" \
+    "${TAG_IMAGE}" \
+    /bin/bash -lc "$script"
+
+  if [ -d dist ]; then
+    mv -f dist "dist-pypi-${TARGETARCH}"
+  fi
+fi

maint/scripts/pypi.manylinux.Dockerfile

+ARG TARGETARCH
 FROM pytorch/manylinux2_28-builder:cuda12.1 AS builder_amd64
 ENV CUDA_VERSION=12.1 \
     AUDITWHEEL_PLAT=manylinux_2_28_x86_64
@@ -6,12 +7,17 @@ RUN pip3 install uv
 FROM pytorch/manylinuxaarch64-builder:cuda12.8 AS builder_arm64
 ENV CUDA_VERSION=12.8 \
     AUDITWHEEL_PLAT=manylinux_2_28_aarch64
+RUN /opt/python/cp312-cp312/bin/pip install uv
 
 FROM builder_${TARGETARCH}
 
 ENV DEBIAN_FRONTEND=noninteractive \
     TZ=Etc/UTC
 
+ENV PATH="/usr/local/cuda/bin:${PATH}"
+
+ENV LD_LIBRARY_PATH="/usr/local/cuda/lib64:${LD_LIBRARY_PATH}"
+
 RUN set -eux; \
     uv venv -p 3.12 --seed /venv; \
     git config --global --add safe.directory '/tilelang'

maint/scripts/pypi_distribution.sh

 set -eux
 
-rm -rf dist
+rm -rf dist raw_dist
 
 python -mpip install -U pip
 python -mpip install -U build wheel auditwheel patchelf

pyproject.toml

@@ -19,7 +19,7 @@ classifiers = [
     "Programming Language :: Python :: 3.12",
     "Intended Audience :: Developers",
     "Intended Audience :: Science/Research",
-    "Scientific/Engineering :: Artificial Intelligence",
+    "Topic :: Scientific/Engineering :: Artificial Intelligence",
 ]
 dynamic = ["version"]
 dependencies = [
@@ -89,7 +89,17 @@ tilelang = "tilelang"
 "tilelang/src" = "src"
 # NOTE: The mapping below places the contents of '3rdparty' inside 'tilelang/3rdparty' in the wheel.
 # This is necessary to find TVM shared libraries at runtime.
-"tilelang/3rdparty" = "3rdparty"
+# Restrict 3rdparty contents in wheel to the same allowlist as sdist
+# TVM
+"tilelang/3rdparty/tvm/src" = "3rdparty/tvm/src"
+"tilelang/3rdparty/tvm/python" = "3rdparty/tvm/python"
+"tilelang/3rdparty/tvm/version.py" = "3rdparty/tvm/version.py"
+# CUTLASS
+"tilelang/3rdparty/cutlass/include" = "3rdparty/cutlass/include"
+"tilelang/3rdparty/cutlass/tools" = "3rdparty/cutlass/tools"
+# Composable Kernel
+"tilelang/3rdparty/composable_kernel/include" = "3rdparty/composable_kernel/include"
+"tilelang/3rdparty/composable_kernel/library" = "3rdparty/composable_kernel/library"
 
 [tool.yapf]
 based_on_style = "yapf"