Re: bsd.own.mk - Re: pkg/44425: devel/lua-alt-getopt install problem

[Robert Elz <kre%munnari.OZ.AU@localhost>]

    Date:        Thu, 20 Jan 2011 22:45:26 +0100
    From:        Hauke Fath <hf%spg.tu-darmstadt.de@localhost>
    Message-ID:  <p0624081cc95e5b84635e@[192.168.10.210]>

  | At 21:52 Uhr +0100 20.01.2011, Joerg Sonnenberger wrote:
  | >The core problem is that you insist on building things inside the NetBSD
  | >src tree, but can't live with the resulting fallout.
  | 
  | I've run that configuration for ~ten years, without any fallout.

I've done that forever (much longer than 10 years) - but I disagree with
Joerg's characterisation: /usr/src is not "the NetBSD src tree" on my
system, it is my source tree, which includes NetBSD sources, GNU sources
(like gcc, groff, ...) X sources (I have /usr/src/xsrc, not /usr/xsrc),
pkgsrc (/usr/src/pkgsrc) and local sources (/usr/src/local) - when I was
at Melbourne, there were sources that came from there as well.

/usr/src (on my systems) is just sources.   I don't really understand the
mentality that suggests that we need to have /usr/foosrc for every new foo
that exists, when /usr/src/foo is much saner - and a much better fit for
the unix filesystem access mode granularity (otherwise every /usr/foosrc needs
to be its own partition, or it must share access permissions with /usr,
which makes no sense at all.)

  | Arguably, <bsd.own.mk> has no business configuring install target related

Yes, the problem here is not pksrc, it is the VERY broken base NetBSD
make system that simply assumes that everything in /usr/src conforms to
its expectations.   It should do that only when it is really building
NetBSD sources, which it cannot tell by a simple test of source directory.

kre