Re: [Systemimager-commits] r3266 - in trunk: initrd_source/make.d make.d

[dann f. <da...@da...>]

On Mon, 2005-11-21 at 05:39 -0600, finley wrote:
> Author: finley
> Date: 2005-11-21 05:39:18 -0600 (Mon, 21 Nov 2005)
> New Revision: 3266
> 
> Modified:
>    trunk/initrd_source/make.d/busybox.rul
>    trunk/initrd_source/make.d/devfsd.rul
>    trunk/initrd_source/make.d/dhclient.rul
>    trunk/initrd_source/make.d/module-init-tools.rul
>    trunk/initrd_source/make.d/rsync.rul
>    trunk/initrd_source/make.d/uclibc.rul
>    trunk/make.d/bc.rul
>    trunk/make.d/discover.rul
>    trunk/make.d/dosfstools.rul
>    trunk/make.d/e2fsprogs.rul
>    trunk/make.d/gzip.rul
>    trunk/make.d/hfsutils.rul
>    trunk/make.d/jfsutils.rul
>    trunk/make.d/kernel.rul
>    trunk/make.d/lvm.rul
>    trunk/make.d/openssh.rul
>    trunk/make.d/openssl.rul
>    trunk/make.d/parted.rul
>    trunk/make.d/pdisk.rul
>    trunk/make.d/popt.rul
>    trunk/make.d/raidtools.rul
>    trunk/make.d/reiserfsprogs.rul
>    trunk/make.d/tar.rul
>    trunk/make.d/util-linux.rul
>    trunk/make.d/xfsprogs.rul
>    trunk/make.d/zlib.rul
> Log:
> Set NCPUS in Makefile.  Use NCPUS for paarrallell make wherever appropriate.  And don't remove that rpm.src.rpm for popt.rul.

Note that doing it recursively like takes scheduling away from make; it
won't be able to limit the number of threads to your -j, so this could
end up running lots many more concurrent tasks than cpus.

Also, note that if you can't do a make -jN at the toplevel (which should
automatically preculate down to each other level), then that is a bug in
our dependency system.