Re: BETA1 5.3 make world DESTDIR failure

On Wed, Aug 25, 2004 at 11:30:02AM -0700, David O'Brien wrote:
> On Wed, Aug 25, 2004 at 08:49:09PM +0300, Ruslan Ermilov wrote:
> > On Wed, Aug 25, 2004 at 10:37:07AM -0700, David O'Brien wrote:
> > > On Wed, Aug 25, 2004 at 04:47:22PM +0300, Ruslan Ermilov wrote:
> > > > as IMAKEENV (from Makefile.inc1) doesn't utilize DESTDIR.  I think I
> > > > know how to fix "make world DESTDIR=/foo" to make it work again -- we
> > > > need to convert all DESTDIR uses in Makefile.inc1 to be command-line
> > > > variables.  I'm currently testing this patch, and it looks promising:
> > > 
> > > Please also test 'make world' without DESTDIR in /etc/make.conf, but with
> > > 1. "HISTORICAL_MAKE_WORLD=yes" in /etc/make.conf
> > > 2. env HISTORICAL_MAKE_WORLD=yes make world
> > > before committing.
> > > 
> > That's pointless for this change.  If HISTORICAL_MAKE_WORLD is defined
> > and DESTDIR is not (irrelevant where), "make world" is just equivalent
> > to "make buildworld" followed by "make installworld".  Clearly, "make
> > buildworld" without DESTDIR works with and without this change.
> 
> I saw some references of "DESTDIR" in your patch and wasn't 100% sure it
> wouldn't be a problem.  I don't understand the resistance to extra
> testing. :-)
> 
It just changes the type of DESTDIR's involved into buildworld from
environment type variables to command-line variables.  Testing a
"make world -DHISTORICAL_MAKE_WORLD" with this change is equivalent
to testing "make buildworld" (without DESTDIR specified anywhere),
and I'm pretty sure this patch didn't touch this codepath.  If you
still want it, could you please test it for me?


Cheers,
-- 
Ruslan Ermilov
ru_at_FreeBSD.org
FreeBSD committer