Re: python31[12] PLIST failure under Linux

[Greg Troxel <gdt%lexort.com@localhost>]

"Morgan, Iain (ARC-TN)[InuTeq, LLC]" <iain.morgan%nasa.gov@localhost> writes:

> I had a somewhat related issue building Python 3.11 from the 2023Q3
> release on two RHEL 8 systems.  In one case, it built and installed
> fine.  In the other, the dbm.so module was built, but not included in
> the PLIST.
>
> On the second system, I merely worked around the issue by adding an
> explicit PLIST.dbm=yes to the pkgsrc Makefile.  However, it sounds
> like the logic for determing whether PLIST.dbm and PLIST.nis should be
> se may need some work.

Or, the package should more carefully control whether dbm is built.
Generally, we try to have package logic decide if it wants something,
and then either

  - if yes, add explicit dependencies to the prereqs are there
  - if no, pass --without-foo to prevent it from being detected

This seems like a middle ground, which IMHO is very likely a bug.