Thomas Deutschmann <whi...@gentoo.org> writes: > the problem is line > >> m4_define_default([am_py_min_ver], m4_ifval([$1], [$1], [2.0])) > > Code like > >> AM_PATH_PYTHON([$PYTHON_VERSION]) > > will result in "am_py_min_ver" set to "$PYTHON_VERSION" which is > obviously not a valid value. I.e. the passed argument isn't expanded > when used in the macro. > > However, I don't understand why we need the dynamic in the list > creation. From my understanding, GNU Pyconfigure, the implementation > source, provided a macro which requires to set lower and upper bound > ("PC_INIT([MIN_VER], [MAX_VER])"). But given that the AM_PATH_PYTHON > public function wasn't changed, we don't need to (better: cannot) > generate the list dynamically.
You are right that the solution chosen by Pyconfigure solves a different issue because of the upper bound. > So I would recommend to change this line to > >> m4_define_default([am_py_min_ver], [2.0]) > > This fixes the problem in cryptsetup, util-linux and PythonMagick for > me. And the check still works. I.e. when I remove the minimum Python > version or newer on my system, configure fails like expected. Given the other portability problems the faulty commit has, I think simply reverting it as suggested by Eric Blake seems better. Thanks. -- Mathieu Lirzin GPG: F2A3 8D7E EB2B 6640 5761 070D 0ADE E100 9460 4D37