Re: [9front] [PATCH] private /srv attach option

[ori@eigenstate.org]

Quoth Jacob Moody <moody@mail.posixcafe.org>:
> 
> > A soup of new features of 'rfork M' quality is a problem;
> > let's figure out where we're going.
> >
> 
> Ouch.
> 
> For the record I did discuss some of these /srv ideas
> in another thread, I believe the it was the one for git/serve to use chdev.
> Nobody seemed interested in discussing it then.
> 

in isolation, this change seems.. fine, I guess. It does
something useful, though it's got some inelegance around
needing a global view to get a clean /srv, which means
that a restricted /srv can't get a clean /srv. It's also
encoded into the interface, so we can't easily fix it
later.

This may or may not matter, I don't know.

the gripe I have is that I know you're aiming turn namespaces
into useful security boundaries, as well as providing other
ways to ratchet down permissions. But I haven't seen a summary
of the changes you feel are needed, the places and ways that
they'd be used, and a diversity of examples that demonstrate
the changes are parsimonious and sufficient.

What is the end goal, and what should it be?

Let's not design in a vaccuum.