On Jun 12, Guenther Grau wrote
> Hi,
> 
> I saw the Manuel Bouyer's implementation of the ext2fs has been commited
> to the tree. I saw the he added an entry to the port-i386/GENERIC config
> file, but to no other config files of other trees. Thus my question:
> Is the ext2fs MD or is it MI?

It is MI (or if it is not it is a bug). I know it compiles on sparc and
alpha, but it has only been tested on i386. So I only added it to
arch/i386/GENERIC.
Perhaps it should be added to other port's GENERIC too, but commented out ?

> If it is MI, then I'd like to repeat an earlier proposal:
> Most of the ports have a std-config file, which is included in the other
> config-files. Can we extend this scheme so that we have a std-file for
> all architectures, which includes all options (commented out or not)
> that are applicable to all ports? Thus a specific port file would look
> like this (don't know the exact config syntax of hand, but you'll get
> the idea):
> 
> #include <std-config>
> #include <std-config-port-your_favorite_port>
> other options ...
> 
> That way things like the new ext2fs could have been added to the
> std-config file and would have been available on all other ports.
> No digging through the other port's config files needed to find out
> it exists and can be used.

EXT2FS is a bad example for that, because I don't think pmax, arm32
or pc532 users want EXT2FS support by default in their kernel.

I don't think this is a good idea. We need a file with all the
config options or devices for a port. If this file is not GENERIC, then
you need to add the new option for all ports, plus <std-config>.
I can't see a win here.

--
Manuel Bouyer, LIP6, Universite Paris VI.                 bouyer@masi.ibp.fr
--