Re: pkg/45450: pkgtools/p5-pkgsrc-Dewey fails to build with non-standard $WRKOBJDIR

[Joerg Sonnenberger <joerg%britannica.bec.de@localhost>]

The following reply was made to PR pkg/45450; it has been noted by GNATS.

From: Joerg Sonnenberger <joerg%britannica.bec.de@localhost>
To: gnats-bugs%NetBSD.org@localhost
Cc: 
Subject: Re: pkg/45450: pkgtools/p5-pkgsrc-Dewey fails to build with
 non-standard $WRKOBJDIR
Date: Wed, 12 Oct 2011 14:26:15 +0200

 On Wed, Oct 12, 2011 at 11:50:05AM +0000, John Marino wrote:
 >  What you interpret as shooting your own foot might be interpreted a bug
 >  in pkgsrc by others.  The directory /usr/obj/ was not selected at
 >  random.  There should be a very good reason why this is off-limits (if
 >  it's really is off-limits), and if so, pkgsrc should be actively
 >  enforcing this limitation as OBATA suggested.
 
 You sound like a cancer patient that tells his doc about the head ache
 he had for years. The very good reason for this being off-limits is that
 it triggers magic behavior in make. Enforcing it is non-trivial as there
 are too many different ways to trigger it.
 
 >  That said, pkgsrc has mostly been working fine literally for months.  I
 >  do notice a "../../pkgtools/digest not found" message at the start of
 >  every build which is probably related to this issue, but other than the
 >  two issues mentioned in this thread, pkgsrc has been behaving well with
 >  the $WRKOBJDIR overridden to the value /usr/obj/usr/pkgsrc.  If a couple
 >  of more relatively links need to be fixed to entirely support this
 >  directory base, we should probably do that rather than passing off these
 >  issues as stupid user error.
 
 Frankly, I am more in favour of something being openly broken that
 seemingly working most of the time. This is such a case where pretty
 much every make target can easily create a more or less obvious error.
 I don't see the advantage in hunting them down.
 
 Joerg