Hi,
Unfortunately, the previous version of the patchset has caused
regressions, so I had to revert the few last commits. Here's a little
different approach that should be safe for existing ebuilds,
and possibly less confusing to ebuild authors.
Moving the ${BUILD_DIR}/install to D merge logic from
distutils-r1_python_install into post-python_install broke ebuilds that
relied on modifying the install tree in python_install. The new version
leaves it in distutils-r1_python_install, so that existing ebuilds
continue working.
The different for 'no' mode is that distutils-r1_python_install needs
to be called if you override python_install and expect the merging
to happen.
Michał Górny (7):
distutils-r1.eclass: Move venv creation to post-compile
distutils-r1.eclass: Future-proof python_install() for empty root
distutils-r1.eclass: Move python_optimize call to post-install
distutils-r1.eclass: Introduce DISTUTILS_USE_PEP517=no mode
dev-python/tomli: Use DISTUTILS_USE_PEP517=no
dev-python/installer: Use DISTUTILS_USE_PEP517=no
dev-python/gpep517: Use DISTUTILS_USE_PEP517=no
dev-python/gpep517/gpep517-6-r1.ebuild | 41 ++++++
.../installer/installer-0.5.1-r1.ebuild | 37 +++++
dev-python/tomli/tomli-2.0.1-r1.ebuild | 36 +++++
eclass/distutils-r1.eclass | 135 ++++++++++++------
4 files changed, 206 insertions(+), 43 deletions(-)
create mode 100644 dev-python/gpep517/gpep517-6-r1.ebuild
create mode 100644 dev-python/installer/installer-0.5.1-r1.ebuild
create mode 100644 dev-python/tomli/tomli-2.0.1-r1.ebuild
--
2.35.1
