Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 20 Jun 2005 12:56:07 -0700
From:      Bakul Shah <bakul@BitBlocks.com>
To:        "M. Warner Losh" <imp@bsdimp.com>
Cc:        rwatson@freebsd.org, current@freebsd.org
Subject:   Re: dhclient less functional with nanobsd because of NO_CXX 
Message-ID:  <200506201956.j5KJu7db084244@gate.bitblocks.com>
In-Reply-To: Your message of "Mon, 20 Jun 2005 12:49:05 MDT." <20050620.124905.35871665.imp@bsdimp.com> 

next in thread | previous in thread | raw e-mail | index | archive | help
> It is generally desirable to have a separate 'install' environemnt
> from the 'build' environment on real embedded systems.  The fact that
> nanobsd doesn't have this useful distinction is a problem with
> nanobsd, not devd.  It should build everything, but install with all
> the NO_XXX flags set to do subsetting.

would it make sense to use an mtree file as a starting point
particularly for embedded systems?  perhaps a tool that takes
output of make -n install + an mtree description of a system
and spits out which install targets need to be built.

the basic problem in my view is that install targets don't
quite fit in the `make' model of maintaining a dependedancy
graph -- make install sends thing *outside* the source tree
that make manages.  the same is true of various other 'magic'
targets since make is being used as a swiss-army-knife,

perhaps mtree should be extended to specify where to pick
things up in order to populate a target tree.  then one can
periodically do something like `mtree --update sys.mtree' and
have it build *eveything* that changed since the last time
you did this operation _and_ save a snapshot so that you can
get back to a previous consistent snapshot.

for what it is worth.




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200506201956.j5KJu7db084244>