Hi, On 2022-01-23 19:49:57 -0800, Andres Freund wrote: > > To avoid too noisy breakages, we could have python.m4 emit INCLUDEPY and > > then > > search the bf logs in a day or three? > > Maybe something like the attached? Not particularly nice, but should give us > most of the relevant information?
FWIW, so far all 73 animals that reported on HEAD that they ran with the change and that currently detect as with_python='yes', find Python.h via INCLUDEPY the same as via get_python_inc(). This includes systems like gadwall where sysconfig.get_path('include') returned the erroneous /usr/local/include/python2.7. Of course that doesn't guarantee in itself that Python.h is usable that way. But none of the systems report a get_python_inc(False) differing from get_python_inc(True), or from the value of INCLUDEPY. So I don't see a reason for why it'd not? Greetings, Andres Freund