Misc ideas & comments
Fri, 27 Mar 1998 10:41:16 -0800 (PST)
>Date: Fri, 27 Mar 1998 09:37:18 -0600
>From: David Tillman <email@example.com>
>Subject: Re: Misc ideas & comments
>> 5. I think that debates like CL versus Scheme are useless if
>> they make the project stop. I don't mind switching to CL if that
>> proves to be easier and more efficient to get LispOS started.
> Absolutely! I love Scheme but I would rather see a Lispy OS
> happen rather than debate Lisp vs Scheme.
But religious arguments can be so much fun, if you don't take it too
> It seems to me that there many things that we could be designing
> *now* before the core is done.
> For example: Assuming that the LispOS is multiuser, what structure
> will we use to store user info.
I think that's a big assumption. One of the really nice things about
the traditional LispMs of old was that they weren't multi user in any
sense that you're talking about. The philosophy was one machine per
user. By making this design decision, one didn't have to worry about
separating and protecting processes from each other. That meant, if
you wanted to hack the GC, go ahead. The only one you'd screw up was
you. If you had multiple users, we couldn't let you because you might
do something that effects the other users.
> We could use the same primitive
> structure as the Unix passwd file. This would seem to be a tragedy
> as we have the opportunity to so muh more.
> An Unixy type would probably look like this:
> ("bob" "N32FSUODn7lo2" 1004 100 "Bob Smith" "MIS Office"
> "x127" 'std-home 'std-shell)
> Perhaps a Lispy version would look like:
> ('bob "N32FSUODn7lo2" 'std-home 'std-shell)
> The symbol 'bob would be used to look up all extra info
> in a LDAP type structure.
I'd think something like this would be more appropriate:
(defclass user (#-POS persistant-base-class)
((name :initarg :name
(home-dir :initarg :directory