--=-=-=
Content-Transfer-Encoding: quoted-printable


"Jeremy C. Reed" <reed@reedmedia.net> writes:

>> reed@reedmedia.net said:
>> > $ cd /opt/rainier/pkgsrc/databases/gramps2
>> > $ make show-var VARNAME=3D_PYTHON_VERSION
>> > 23=20
>>=20
>> Joerg Sonnenberger has set PYTHON_VERSIONS_ACCEPTED
>> for gramps to 24 only, so that shouldn't happen anymore,
>> independently of the default setting.
>
> I didn't notice that and it was still broken. I had already updated=20
> this a couple times since he did that.
>
> $ make show-var VARNAME=3D_PYTHON_VERSION=20
> 23
>
> $ ident Makefile=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=
=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=
=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20
> Makefile:
>      $NetBSD: Makefile,v 1.19 2006/07/12 15:27:18 joerg Exp $
>
> $ make show-var VARNAME=3DPYTHON_VERSIONS_ACCEPTED
> 24 23 22
>
> $ pwd
> /opt/rainier/pkgsrc/databases/gramps2
>
> I see this is caused by including ../../databases/py-bsddb3/buildlink3.mk=
=20
> which has (since 2005):
>
> PYTHON_VERSIONS_ACCEPTED=3D       24 23 22
>
> which conflicts with this.

I suspect the change of the default to 24 will largely avoid these
issues, at least until packages start needing 25.  But, it seems that
we really need a set intersection operation so we can do

PYTHON_VERSIONS_ACCEPTED &=3D 24 23 22

and that intersects the current variable with the list (perhaps
treating the empty variable as the universe and hence being like =3D).

Perhaps we could use guile to extend make :-)

=2D-=20
    Greg Troxel <gdt@ir.bbn.com>

--=-=-=
Content-Type: application/pgp-signature

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.5 (NetBSD)

iD8DBQFE2JKj+vesoDJhHiURAgPoAJ9cEXwDu1vi3jqh4HpoicR+F9l5+QCfasmJ
KCPzkoSmQwKq80JxMHQapDw=
=usQb
-----END PGP SIGNATURE-----
--=-=-=--