tech-pkg archive
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]
Re: cwrappers now enabled by default on some platforms
On Mon, Oct 31, 2016 at 07:06:04PM -0400, Greg Troxel wrote:
>
> Joerg Sonnenberger <joerg%bec.de@localhost> writes:
>
> > On Mon, Oct 31, 2016 at 04:47:04PM +0000, David Holland wrote:
> >> On Fri, Oct 28, 2016 at 11:39:06AM +0100, Robert Swindells wrote:
> >> > >That is a killer, any details?
> >> >
> >> > ctfconvert -g -L VERSION -g alloc.o
> >> > make: exec(ctfconvert) failed (No such file or directory)
> >>
> >> shouldn't be trying to run that without probing for it...
> >
> > Inconsistent build. cwrappers is using plain bsd.prog.mk, but if the
> > system was build with MKCTF=no, but /etc/mk.conf doesn't contain it,
> > things will naturally break...
>
> I see your point, but if one does a build with some mk.conf, and gets
> binaries, and later installs them, I don't think we generally have rule
> that mk.conf must contain all the settings. pkgsrc should be treating
> the base system as something that's installed, not something we build.
No, it is rather obvious that you do *not* understand my point.
The cwrappers package is using a simple BSD Makefile. No more, no less.
We generally have no policy for dealing with crippled NetBSD
installations. So if someone want to insist that the base system is
crippled, it is the responsibility of that person to provide consist
mk.conf settings. I can't reproduce this problem on netbsd-7-0 at least,
which defaults to MKCTF=no. I can't currently check my current system,
since it is still busy with other things. In short, we treat the base
system as a consistent entity and providing a working make is part of
that. If you don't like that, you can bootstrap. bootstrap-mk-files has
no knowledge about ctfconvert.
> So I agree with dholland that there should be some autoconf-like
> probing.
It doesn't. Look into /usr/share/mk for the logic it uses, but it is
essentially controlled by MKCTF. Why it doesn't work for OP -- no idea.
> A harder question is if changing any MK setting on the base system
> should require the equivalent of rebootstrapping. I think it would be
> much better if it didn't.
That's a completely unrelated and irrelevant discussion at this point.
Joerg
Home |
Main Index |
Thread Index |
Old Index