On Tue, 3 Apr 2001 wojtek@wojtek.from.pl wrote:

> probably. 
> > Yes, people pointed out other OS have ACLs. But that's not the reason we'd
> 
> i've read that ACL should be implemented because new samba make use of it
> and it make file serving for windoze users better etc.etc..
> 
> that made me worried very much.

That's why the first person asked for them. That's not the only (or even
the main) reason we want to do them.

> > add ACLs to NetBSD. The reason was in the thread - we'd add it because
> > people have used ACLs on other OSs AND FOUND THEM USEFUL. ACLs' utility is
> > why we'd add them to NetBSD.
> > > it's name is linux!
> > 
> > Linux supports sound on my Beige G3 Powermac. So are you saying that if I
> > want sound on it, I should switch to Linux? That's where your arguement
> 
> no. you should try to write driver for netbsd :)
> 
> i'm talking about software, not hardware

So I shouldn't have done the layered file system work I did? That's
software only. We shouldn't have new file systems, like NTFS? We shouldn't
have a unified buffer cache? We shouldn't have scheduler activiations? We
shouldn't tune the kernel for better performance? All of these things are
software.

> > would lead me. Do you not see how restrictive that arguement is? Because
> > you're basing the arguement on what Linux does rather than the quality of
> > an inovation, the arguement stifles all inovation.
> no

Then please look again. You've changed the arguement by allowing hardware
drivers which overlap Linux, but your, "I'm talking about software,"
comment indicates that you're still trying to restrict software features
which mirror those of Linux. That's still restrictive.

Yes, just because Linux does it is a bad arguement (unless we're talking
about Linux emulation code). But that shouldn't mean that something we
think is good gets rejected because Linux happened to do it before we did.

> > I think it's great that the permissions system works excelently for you.
> > But are you the only user of NetBSD? No (since I am a NetBSD user and I am
> > not you I know the answer's no :-) . So how do you know that just because
> > you don't need ACLs, no one else does? Phil Nelson pointed out a case
> > where ACLs would have been VERY useful. Other folks noted where they would
> > find ACLs useful. So ACLs have utility to NetBSD users.
> 
> how difficult is to use /etc/groups more intensively?
> it's really ease.

Did you read Phil's notes? How do you put yourself in 17 groups at once?

> > > please make already good UNIX better even more instead of bloating with
> > > "very much needed features".
> > > 
> > > making netbsd popular OS is important, but making it good, smart and
> > > proffesional (in good meaning) is far more important.
> > 
> > And from having worked at and talked with coleagues from high-performance
> > computing centers, ACLs (if done right) will make NetBSD a much more
> > professional OS.
> 
> no. i do not mean "professional" as "windows 2000 professional" means.

Where have I mentioned Windows 2000?

> > > popularity != quality.
> > 
> > Agreed. But as I said (many times) above, other-OS feature lists aren't
> > why we're interested in ACLs, the fact we find them useful is. :-)
> 
> maybe i'm wrong. could you point me to some page about ACL (about idea,
> not windoze centric). maybe i could educate more.

Ahh. If you thought that ACLs came from Windows, then yes, I can see your
confusion. ACLs, as I think Bill Sommerfeld mentioned, are older than
UNIX. I'm not really sure what the best ACL reference is. But try to find
out some on permissions for AFS, or the security part of DCE. I hope other
folks will contribute references too. These are all very UNIX-centric ACL
environments.

Take care,

Bill