Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 23 Mar 1995 12:31:04 +0100 (MEZ)
From:      me@dude.pcs.dec.com ( Michael Elbel )
To:        kuku@gilberto.physik.rwth-aachen.de
Cc:        freebsd-current@freefall.cdrom.com
Subject:   Re: Make World Times and a question about shared libs / make all
Message-ID:  <m0rrl6Y-0005fCC@dude.pcs.dec.com>
In-Reply-To: <199503230930.KAA14532@gilberto.physik.rwth-aachen.de> from "Christoph Kukulies" at Mar 23, 95 10:30:54 am

next in thread | previous in thread | raw e-mail | index | archive | help
> > Now I've noticed that subsequent 'make all install's will relink
> > all the programs. I suppose this is due to the libraries being
> > newer than the existing objects. However, this wouldn't be necessary
> > unless the shared lib version has gone up. How hard would it be
> > to get rid of this behaviour? I don't want to go and dig myself
> > up to the armpits into the make stuff if it's impossible :-)
> 
> I'm not quite sure what you mean. Do you want to make  make all install
> faster when a make world has been done already and all the objects
> sit still there in the tree? I think this is not what we want.

Why not? After all I'm *not* doing a make world again. I'd like it
to run as fast as possible. And yes, I know that sometimes there
are things changing which will require careful thought or a make world,
for this I'm reading cvs-all :-)

> Make world (which includes cleandir or is it clean?) is slow because
> the libs are being built several times during the course of makeing the world.
> 

Again, I'm not doing a make world. Ideally I'd like the rule to generate 
shared linked binaries not to check on the date of the shared library but 
on its version, since the interfaces will stay the same and any changes
be automaticaly picked up. I dimly remember the exact same thing having
been discussed here.

The actual dependencies could be on some sort of cookie in the obj tree
that only gets touched when the major number of the library changes.
The tricky part then would be to get make depend to do this right :-)

Michael
--
Michael Elbel, Digital-PCS GmbH, Muenchen, Germany - me@FreeBSD.org
Fermentation fault (coors dumped)



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?m0rrl6Y-0005fCC>