On 23 May 2011, at 22:29, Jeppe Toustrup wrote:

> 2011/5/23 Ken Gunderson <[email protected]>:
>> On Mon, 2011-05-23 at 15:39 -0400, Alex Smith (K4RNT) wrote:
>>> Another related question - why have we stopped using pfexec and
>>> started using sudo? I preferred RBAC...
>> 
>> Have we?  I've been testing 148b and just assumed it was a defect.  If
>> not, I concur with you that RBAC is preferrable to sudo.
> 
> The change was made upstream. See this bug report which discusses the change:
> https://defect.opensolaris.org/bz/show_bug.cgi?id=4885

This looks to me like a comedy of analogical errors: it's not, for example, 
that Windows users no longer have access to admin privileges, it's just that 
asserting those privileges requires authentication. Similarly, it's not that 
sudo is the right way to do things in general, nor is it all that important 
that using sudo provides an affordance to people accustomed to using it 
elsewhere. It's that pfexec doesn't have a mechanism to require authentication 
for the assertion of particular levels of privilege, and it looks like a 
decision was made to kick the can down the road on that. You could follow the 
various offered analogies conclude not that access to root should be mediated 
by sudo because that's what people expect but that people rightly expect there 
to be an authentication requirement, in principle if not in fact, between them 
and access to those privileges. It wouldn't be a show-stopper to say that for 
most people's purposes, pfexec is like sudo, it's just called pfexec and has a 
different configuration system because the privileges have a different 
structure, if you need to edit the config files, read the fine man page. The 
problem is a decision was made, however implicitly or explicitly, not to fill 
in the functional gap and add authentication. Instead of agreeing that pfexec 
needed to be like sudo in this respect and making it that way, we just got sudo.

That continues to be the case because the determinative constraints haven't 
moved an inch. I don't see the point in the subsequent Talmudic arguments about 
references to Ubuntu, as the protracted argument doesn't add up to the 
inconsequence of the point.

> --
> Venlig hilsen / Kind regards
> Jeppe Toustrup (aka. Tenzer)
> 
> _______________________________________________
> OpenIndiana-discuss mailing list
> [email protected]
> http://openindiana.org/mailman/listinfo/openindiana-discuss

Attachment: PGP.sig
Description: This is a digitally signed message part

_______________________________________________
OpenIndiana-discuss mailing list
[email protected]
http://openindiana.org/mailman/listinfo/openindiana-discuss

Reply via email to