If we're going to support FFS_EI it makes sense the userland data files should
also try and be as transparent as possible. (I realize in the general case
these 2 things aren't the same but the idea is still the same. I should be
able to plug disks in from other boxes and it would be just as nice to take
the local data files and use then to say rsync over to other boxes).

James

>
>
>>> 	is it possible for us to put a fake entry (like "*endian*"), which
>>> 	includes endian indication, into pwd.db?
>>> 	- if there's an entry, password database lookup should convert endian
>>> 	  as necessary.
>>> 	- if there's no entry (= pwd_mkdb is old and library is new),
>>> 	  too bad, it will be just like we are now.
>>> 	- if pwd_mkdb is newer, fake entry should be harmless.
>
>	an additional note: i'm still not 100% sure if it is the way to go.
>	it sounds too much, if /etc is supposed to be cpu/arch dependent
>	endian.
>	we definitely need target-endian indication in builds anyways, for
>	other places too (to build /usr/libdata and other target-endian data
>	generation), and if we have the indication, we'll be able to use that
>	for pwd.db generation and we will be happy.
>
>	so my question is, do we support copying pwd.db across machines,
>	specifically with different endian?
>
>itojun
>
>
>
>