re: toolchain/52834: NetBSD/sparc cross-build failure

[matthew green <mrg%eterna.com.au@localhost>]

The following reply was made to PR toolchain/52834; it has been noted by GNATS.

From: matthew green <mrg%eterna.com.au@localhost>
To: clare%csel.org@localhost
Cc: Robert Elz <kre%munnari.OZ.AU@localhost>, toolchain-manager%netbsd.org@localhost,
    gnats-admin%netbsd.org@localhost, netbsd-bugs%netbsd.org@localhost,
    gnats-bugs%NetBSD.org@localhost
Subject: re: toolchain/52834: NetBSD/sparc cross-build failure
Date: Sat, 06 Jan 2018 07:56:59 +1100

 clare%csel.org@localhost writes:
 > On Sun, 17 Dec 2017 15:45:01 +0000 (UTC)
 > Robert Elz <kre%munnari.OZ.AU@localhost> wrote:
 > 
 > >    | i thought OBJMACHINE="yes" was sufficient to separation of objdir,
 > >  
 > >  I think that might happen if you use -M rather than -O to set the objdir
 > >  location.   I prefer to have more control over where things are put, so
 > >  I have never used that method...
 > 
 > both -M and -O itself does not separate objdir per $machine for me,
 > but build problem was solved by fixing scripts...
 
 it can -- as long as you include the right stuff in the value.
 it's a make variable, so it can be quite complex.  i use this:
 
    /var/obj/${MACHINE}${MACHINE_ARCH:N${MACHINE}:C/(.)/-\1/}
 
 which gives me eg /var/obj/sparc, etc, for MACHINE=MACHINE_ARCH,
 and eg /var/obj/amd64-x86_64 for others.  i have a variant that
 tries to guess gcc vs clang, but ACTIVE_CC isn't set early
 enough in all cases.  (i am an MAKEOBJDIRPREFIX user, fwiw.)
 
 i also try to *always* build from a read-only mount of the srcs
 which avoids or clearly fails many types of issues.
 
 
 .mrg.