Subject: Re: Support for ACLs
To: Todd Vierling <tv@wasabisystems.com>
From: Lord Isildur <mrfusion@umbar.vaxpower.org>
List: tech-kern
Date: 03/08/2001 12:41:23
> Please come back when you have a real technical reason for why ACLs will be
> detrimental to ffs and/or NetBSD as a whole.
> 

It is cruft that is too specialized to be put in all the FFS code. 
I pretty categorically like to crusade against cruft. 
However, as long as a non-AClified-FFS supoprting kernel can mount and use
an ACL-ified FFS silesystem, and an ACLified kernel can mount and use
a non-ACLified FFS without ACLifying it in the process, then it will be ok-
it will mean that when i run into an ACLified FFS i can read with with my
systems, none of which will have any ACLified FFS code on them, and when i
encounter an ACLified system it can read my non-ACLified FFS filesystems. 
Though it often makes me sound like a broken record, and might get on 
peoples' nerves, i do not see NetBSD as a popularity contest. I don't mind
that Linux is more popular- i rather like it, because it keeps most of 
the (l)users out of the NetBSD community , where they'd ruin it as fast as
linux went down the toilet in '95. I tend to be a purist, and to voice 
arguments of that nature. NetBSD has meen imho consistently the superior 
contemporary UNIX. Part of this was that we had a must stronger 
resistance to the creepy feature creature in previous years, and concentrated
on supporting every hardware known to man and on performance instead of 
features. Now there is an incredible pressure to pour in a kitchen sink's 
worth of features. My ravings are just a voice to say ' beware the feeping
creatures!'. I say BSD is _better_ and should _not_ keep trying to be
just like the other UNIXen, which are imho inferior and crappy. If it means
less 'compatibility' , so be it. Better to be better and incompatible 
than to stoop and roll in the mud with the other crappy systems. There is no
single feature that makes this happen; it's the slow creep of _all_ of them
that integrates to that result. 
Anyhow, as i mentioned, as long as there is bidirectional 
interoperability between ACLified and nonACLified FFS filesystem code and 
filesystems, with _no_ modifications to the non-ACLified code, then this
shouldnt be such a problem. Folks like me can not compile it into our 
kernels, and the filesystems and machines on both sides of the divide 
will be able to still exchange filesysems. It'll mean that ACLs will be an
optiional feature even in an ACLified system: this shoudl not be a 
problem. 

Isildur