Skip to content

v0.3.0

Latest

Choose a tag to compare

@github-actions github-actions released this 17 Feb 00:03
· 9 commits to main since this release
e10b718

Bzlmod

bazel_dep(name = "rules_cc", version = "{rules_cc_version}")
bazel_dep(name = "rules_cuda", version = "v0.3.0")

cuda = use_extension("@rules_cuda//cuda:extensions.bzl", "toolchain")
cuda.toolkit(name = "cuda")
use_repo(cuda, "cuda")
Legacy WORKSPACE config
load("@bazel_tools//tools/build_defs/repo:http.bzl", "http_archive")

http_archive(
    name = "rules_cc",
    sha256 = "{rules_cc_sha256}",
    strip_prefix = "rules_cc-{rules_cc_version}",
    urls = ["https://github.com/bazelbuild/rules_cc/releases/download/{rules_cc_version}/rules_cc-{rules_cc_version}.tar.gz"],
)
load("@rules_cc//cc:extensions.bzl", "compatibility_proxy_repo")
compatibility_proxy_repo()

http_archive(
    name = "rules_cuda",
    sha256 = "e677c40b88f77843587f5b51667ecd402cc7e813fcf879996a9ac0c369e0df24",
    strip_prefix = "rules_cuda-v0.3.0",
    urls = ["https://github.com/bazel-contrib/rules_cuda/releases/download/v0.3.0/rules_cuda-v0.3.0.tar.gz"],
)

load("@rules_cuda//cuda:repositories.bzl", "rules_cuda_dependencies", "rules_cuda_toolchains")
rules_cuda_dependencies()
rules_cuda_toolchains(register_toolchains = True)

What's Changed

  • refactor!: change for hermetic with minor user interface breaking by @cloudhan in #300
  • feat: hermetic ctk with deliverable by @cloudhan in #285
  • test: test components for both bzlmod and workspace by @cloudhan in #329
  • feat: load("@rules_cuda_redist_json//:redist.bzl", "rules_cuda_components") by @cloudhan in #286
  • refactor: remove use_cpp_toolchain by @udaya2899 in #330
  • Fix docstring typo by @keith in #341
  • Fix typo in archive_override example by @keith in #340
  • Add CcInfo external_includes by @keith in #342
  • feat: generate version.json to <component_name>/version.json by @cloudhan in #336
  • fix(nvcc): append to cpp_compile PATH by @lukasoyen in #345
  • refactor!: rename @local_cuda as @cuda and @local_cuda_ as @cuda_ by @cloudhan in #343
  • feat: automatically detect compiler_executable from registered cc_toolchain by @udaya2899 in #331
  • Update deprecated CI runners by @freeform-andre in #356
  • fix: allow override opt dbg and fastbuild flags by @cerisier in #363
  • fix: minor fix for cuda_binary and cuda_test macros attrs forwarding by @cloudhan in #365
  • minor bug fix where rename attrs was not being expanded properly by @AlexandreAmice in #366
  • feat: hermetic CUDA clang support by @udaya2899 in #360
  • fix: fix clang detection on windows by @cloudhan in #369
  • feat: optionally passthrough flags from cpp fragments by @cloudhan in #353
  • Rename duplicate cupti_header target by @lgulich in #370
  • Shorten RUNPATH when linking to versioned shared libraries by @ryabkovdv in #371
  • refactor!: rework hermetic clang by @cloudhan in #375
  • refactor: remove components_mapping_compat.attr to stop resolving label during analysis time by @cloudhan in #378
  • Import stubs/*.so as interface libraries by @ryabkovdv in #380
  • feat: implement redist_json for bzlmod by @cloudhan in #379
  • fix: also add additional_header_deps for @cuda//:cuda by @cloudhan in #381
  • feat: add more new archs by @cloudhan in #385
  • style: avoid hardcoding typename of type(object) to avoid potential future break by @cloudhan in #384
  • refactor: make tpl gen for different local ctk, deliverable ctk and deliverable component straightforward by @cloudhan in #386
  • If no components specified, use all components. by @finn-ball in #382
  • Export all files from component repo by @finn-ball in #389
  • feat: search for the sysroot in compile args in toolchain rule by @cloudhan in #390
  • feat: allow specify archs at per-target level by @cloudhan in #394
  • feat: improve build reproducibility by removing --generate-line-info by @kirschem-fernride in #406
  • fix: make cuda_redist_json components attribute optional by @cloudhan in #411
  • fix: add nvtx3 by @cloudhan in #412
  • fix: add libcudacxx under cccl by @cloudhan in #413
  • fix: auto select all components that is available on the platform by @cloudhan in #414
  • feat: add support for CTK 13 by @cloudhan in #410
  • Improve clang toolchain discovery, and support hermetic toolchains that don't use built_in_include_directories by @calebzulawski in #421
  • feat: add nvcc_fixed_random_seed feature for nvcc based toolchain config by @cloudhan in #374
  • Retain cc toolchain flags by @calebzulawski in #419
  • add .inc to the allowed cuda hdrs by @AlexandreAmice in #424
  • fix: import rules_cc components explicitly by @wep21 in #423
  • ci: use bazel 9 lts by @wep21 in #429
  • Add empty queryable cuda_runtime target for disabled cuda by @lalten in #405
  • feat: support external_include_paths feature in nvcc toolchain by @lalten in #430
  • refactor: unify paths collection logic in toolchain configs by @cloudhan in #431
  • feat: add ptxas to label by @cloudhan in #432
  • fix: fix clang rdc compilation by @cloudhan in #435
  • fix: make clang recognize nvvm repo as valid cuda installation by @cloudhan in #433
  • fix: dummy nvcc resolution by @cloudhan in #446
  • Update docs for v0.3.0 by @cloudhan in #447
  • fix: load more rules_cc for bazel 9.x by @wep21 in #450

New Contributors

Full Changelog: v0.2.5...v0.3.0