On Sun, 2021-06-20 at 14:48 +0200, Ulrich Mueller wrote: > > > > > > On Sun, 20 Jun 2021, Michał Górny wrote: > > > @@ -712,8 +714,13 @@ python_newexe() { > > ) > > > > # install the wrapper > > - _python_ln_rel "${ED%/}"/usr/lib/python-exec/python-exec2 \ > > - "${ED%/}/${wrapd}/${newfn}" || die > > + if [[ ${EAPI} != [67] ]]; then > > + dosym -r /usr/lib/python-exec/python-exec2 \ > > + "${wrapd}/${newfn}" > > + else > > + _python_ln_rel "${ED%/}"/usr/lib/python- > > exec/python-exec2 \ > > + "${ED%/}/${wrapd}/${newfn}" || die > > + fi > > Wouldn't it be cleaner to inherit eapi8-dosym.eclass in EAPIs 6 and 7 > instead? Then dosym -r could be used unconditionally, and function > _python_ln_rel could be dropped altogether? > > Same for python-r1.eclass (patch 20/28) and distutils-r1.eclass > (patch 27/28).
Hmm, I suppose that makes sense. Thanks! -- Best regards, Michał Górny