From owner-freebsd-current  Wed Apr 10 10:24:21 1996
Return-Path: owner-current
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.3/8.7.3) id KAA23596
          for current-outgoing; Wed, 10 Apr 1996 10:24:21 -0700 (PDT)
Received: from GndRsh.aac.dev.com (GndRsh.aac.dev.com [198.145.92.241])
          by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id KAA23588
          for <current@FreeBSD.ORG>; Wed, 10 Apr 1996 10:24:16 -0700 (PDT)
Received: (from rgrimes@localhost) by GndRsh.aac.dev.com (8.6.12/8.6.12) id KAA16181; Wed, 10 Apr 1996 10:23:43 -0700
From: "Rodney W. Grimes" <rgrimes@GndRsh.aac.dev.com>
Message-Id: <199604101723.KAA16181@GndRsh.aac.dev.com>
Subject: Re: speedup idea for 'make world'
To: nate@sri.MT.net (Nate Williams)
Date: Wed, 10 Apr 1996 10:23:43 -0700 (PDT)
Cc: peter@jhome.DIALix.COM, current@FreeBSD.ORG
In-Reply-To: <199604100652.AAA00758@rocky.sri.MT.net> from Nate Williams at "Apr 10, 96 00:52:27 am"
X-Mailer: ELM [version 2.4ME+ PL11 (25)]
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: owner-current@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk

> > I'm sitting watching a make world crawling along on a remote machine at the
> > moment, and had an idea..
> > 
> > When it's bootstrapping everything by building the initial tools and
> > libraries, and installing them so it can delete and rebuild everything, it's
> > wasting an awful lot of time building and installing man pages on the first
> > pass.
> > 
> > Perhaps all the:
> >   cd ${foo}/where/ever && ${MAKE} depend all install ${whatever}
> > 
> > should be:
> >   cd ${foo}/where/ever && ${MAKE} -DNOMAN=foo depend all install ${whatever}
> > ?
> > 
> > This would cut out the gzip pass on the libc and other library man pages, and
> > the extremely slow linking while installing.    Especially since it's going to
> > to it again and the man pages are not going to be needed for the rest of the
> > build sequence..

I agree with the above, that would be a good optimization...

> Another thing.  Why are we bothering with the depend step when we're
> going to end up blowing away the stuff anyway.
As long as the blowing away really occurs this would be okay, but there
are now knobs that prevent this, and in that case this could cause problems.

> Especially considering
> that 'make depend all' doesn't work like it implies.  As has been said,
> if we want to actually use the .depend file, it should read 'make depend
> && make all'.

This should probably be fixed, though in most of the cases it does work
correctly because these are being run on subdir levels and the order of
doing things in those cases is correct.

> More thoughts...

Make world was written 3 years ago, and has been hacked on ever since,
probably time to throw the baby and the bathwater out and start over
from scratch...

-- 
Rod Grimes                                      rgrimes@gndrsh.aac.dev.com
Accurate Automation Company                 Reliable computers for FreeBSD