//python/cc:py_cc_toolchain_info.bzl

Provider for C/C++ information from the toolchain.

See also

• Custom toolchains for how to define custom toolchains.

• py_cc_toolchain rule for defining the toolchain.

provider PyCcToolchainInfo

C/C++ information about the Python runtime.

PyCcToolchainInfo.<init>(headers, headers_abi3, libs, python_version)

PyCcToolchainInfo.headers: struct

Information about the header files, struct with fields:

• providers_map: a dict of string to provider instances. The key should be a fully qualified name (e.g. @rules_foo//bar:baz.bzl#MyInfo) of the provider to uniquely identify its type.

  The following keys are always present:

  • CcInfo: the CcInfo provider instance for the headers.

  • DefaultInfo: the DefaultInfo provider instance for the headers.

  A map is used to allow additional providers from the originating headers target (typically a cc_library) to be propagated to consumers (directly exposing a Target object can cause memory issues and is an anti-pattern).

  When consuming this map, it’s suggested to use providers_map.values() to return all providers; or copy the map and filter out or replace keys as appropriate. Note that any keys beginning with _ (underscore) are considered private and should be forward along as-is (this better allows e.g. :current_py_cc_headers to act as the underlying headers target it represents).

PyCcToolchainInfo.headers_abi3: struct | None

If available, information about ABI3 (stable ABI) header files, struct with fields:

• providers_map: a dict of string to provider instances. The key should be a fully qualified name (e.g. @rules_foo//bar:baz.bzl#MyInfo) of the provider to uniquely identify its type.

  The following keys are always present:

  • CcInfo: the CcInfo provider instance for the headers.

  • DefaultInfo: the DefaultInfo provider instance for the headers.

  A map is used to allow additional providers from the originating headers target (typically a cc_library) to be propagated to consumers (directly exposing a Target object can cause memory issues and is an anti-pattern).

  When consuming this map, it’s suggested to use providers_map.values() to return all providers; or copy the map and filter out or replace keys as appropriate. Note that any keys beginning with _ (underscore) are considered private and should be forward along as-is (this better allows e.g. :current_py_cc_headers to act as the underlying headers target it represents).

Added in version 1.7.0: The features.headers_abi3 attribute can be used to detect if this attribute is available or not.

PyCcToolchainInfo.libs: struct | None

If available, information about C libraries, struct with fields:

• providers_map: A dict of string to provider instances. The key should be a fully qualified name (e.g. @rules_foo//bar:baz.bzl#MyInfo) of the provider to uniquely identify its type.

  The following keys are always present:

  • CcInfo: the CcInfo provider instance for the libraries.

  • DefaultInfo: the DefaultInfo provider instance for the headers.

  A map is used to allow additional providers from the originating libraries target (typically a cc_library) to be propagated to consumers (directly exposing a Target object can cause memory issues and is an anti-pattern).

  When consuming this map, it’s suggested to use providers_map.values() to return all providers; or copy the map and filter out or replace keys as appropriate. Note that any keys beginning with _ (underscore) are considered private and should be forward along as-is (this better allows e.g. :current_py_cc_headers to act as the underlying headers target it represents).

PyCcToolchainInfo.python_version: str

The Python Major.Minor version.