diff options
| author | mikari <[email protected]> | 2026-07-07 20:01:46 +0300 |
|---|---|---|
| committer | mikari <[email protected]> | 2026-07-07 20:26:42 +0300 |
| commit | 79f6d536c89e93956e68921a220de82be0314b51 (patch) | |
| tree | 8920ca1b7188bc1160d3abad25ddbdd116f71613 /contrib/python/python-dateutil/py3/dateutil/parser/__init__.py | |
| parent | e8319c1f69486be12a132355af5b30be69a5bc20 (diff) | |
Add CMAKE_EXPORTED_TARGET_NAME_FROM_PATH macro, fix cmake target artefact rename rules-
A project may contain many same-named programs: for example, each integration
test directory defines its own PROGRAM() conventionally called `pipeline`.
CMake requires target names to be unique across the whole project, so exporting
such a subtree to opensource forces one to write
CMAKE_EXPORTED_TARGET_NAME(unique-name) into every such module. The natural
unique name is the module path, but hardcoding path-derived names in every
ya.make is verbose and breaks on every repository move.
This PR adds CMAKE_EXPORTED_TARGET_NAME_FROM_PATH(project/root): it derives the
exported cmake target name from the module path automatically, cutting off the
irrelevant common prefix (and validating that the module really lives under it).
The output artefact keeps its short name, so tests and tooling that run the
binary are unaffected.
It also makes the underlying OUTPUT_NAME rename rules actually work in sem
export: a multi-command variable value substituted in argument position of a
sem string fails to render (`unexpected multicommand substitution`), so the
variable is moved into its own command-position slot.
commit_hash:aa970726f25773150cc839906aaddc98ee8ac412
Diffstat (limited to 'contrib/python/python-dateutil/py3/dateutil/parser/__init__.py')
0 files changed, 0 insertions, 0 deletions
