Re: py-importlib-metadata build problem on 9.0-RC1

[Mike Pumford <mpumford%mudcovered.org.uk@localhost>]

On 06/12/2019 22:41, Roland Illig wrote:
> Am 06.12.2019 um 22:58 schrieb Mike Pumford:
> > Getting the following error from my pkgcomp NetBSD 9.0-RC1 package builds:
> >
> > => Checking file permissions in py37-importlib-metadata-1.2.0
> > error:
> > /pkg_comp/obj/pkgsrc/devel/py-importlib-metadata/default/.../changelog.rst:
> > world-writable file.
>
> devel/py-importlib-metadata/Makefile has been changed several times in
> the last few days. It probably differs between your two systems.
Both builds were done from the same pkgsrc checkout as it is shared 
between the chroots for the 9.0-RC1 and 8.1-STABLE builds.

> In Makefile 1.8 and 1.9 the file permissions were fixed automatically.
> In Makefile 1.10 that fix got removed, without any helpful commit
> message. The DISTFILE of that package still contains world-writable files.
Looking at my checkout I have revision 1.11 of that Makefile for both 
builds so its very odd that 8.1 doesn't complain about permissions 
issues. Perhaps that is why the someone thought the fixup could be 
removed from the pkgsrc makefile?

I'll patch that post-extract back into the Makefile for now :)

> Maybe it is even a good idea to fix the pkgsrc infrastructure so that
> the file permissions are fixed automatically right after post-extract,
> to prevent this class of mistakes from causing any manual work. We
> wouldn't notice any upstream packaging mistakes anymore, and maybe we
> just shouldn't care.
On the surface that seems like a good idea but I think it might be 
difficult in practice. It would be very hard to come up with a set of 
rules that wouldn't screw up permissions sometimes.

Mike