[cap-talk] Yes., PSOS and capabilities, coal and charcoal

[Jed Donnelley]

On 4/10/2008 3:36 PM, Jonathan S. Shapiro wrote:
> On Thu, 2008-04-10 at 15:21 -0700, Jed Donnelley wrote:
>> Butler's ... own "serious" work on capabilities that
>> seem to me mostly the failed CalTSS system.
> 
> I have heard the thought expressed by more than one person that Butler
> so adamantly dislikes capabilities mainly because he couldn't make them
> work. I know who originated this meme, but I won't out them here. The
> amusing surprise was that it was a highly contagious meme.

Contagious and perhaps likely to self generate from the available
input.  I don't believe I'd heard that thought elsewhere before
it occurred to me.

> Personally, I think this isn't fair. Cal/TSS failed for pretty clear
> reasons that had almost nothing to do with capabilities...

I agree.  All the issues we had with NLTSS had nothing really
to do with capabilities - except perhaps some of the overhead
issues we dealt with regarding exchanges which were due to
"small" domains that one might consider to be induced by
capabilities or any alternative POLA mechanism.

>> Incidentally, I'm confident Butler includes
>> PSOS in that 'tried and failed' category - as I believe he
>> rightly should.
> 
> I'm not aware that PSOS was ever deployed. I can ask Peter if anybody is
> interested. At least one of its successors (ASOS, KSOS) *was* deployed.
> That's as good a measure for a research success in this area as one can
> imagine, so I don't agree that PSOS was a failure.

Hmmm.  Perhaps I have the wrong reference.  I just looked through
PSOS Revisited:

http://www.csl.sri.com/users/neumann/psos03.pdf

and couldn't find what I was looking for.  I thought PSOS was
the system that started out using a capability base and then
abandoned it in response at least partly to the concerns
of Boebert et. al. r.e. MLS.  Even if that was PSOS (I think
not now) I wouldn't consider PSOS a 'failure' (a loaded term
in any case - particularly for a research system), but it could
be suggestive of the inability of the capability paradigm to
support needed semantics (e.g. MLS - yes, yes, I know now
disproved).

However, I see that I either misremembered or misplaced that
thought.  If that is what happened with PSOS it seems it
surely would have shown up in that retrospective.  Forget my
comment above for now and I'll see if I can pick up that
thread after some further research.  I apologize if I
promulgated inaccurate information above.  I notice I
suggested this thought at least once before, e.g. it
shows up in:

http://article.gmane.org/gmane.comp.capabilities.general/7178

as: "1973: PSOS - SRI ******* (abandoned capabilities)"
<another likely mistaken reference from me>

Incidentally, how many places is the cap-talk list mirrored?
I often search for phrases that I suspect exist in archived
cap-talk messages and find them in odd places like the above.
The above seems to be a copy of:

http://www.eros-os.org/pipermail/cap-talk/2007-August/008724.html

>> Dr. Sleep found it difficult to
>> believe that I'd never before seen coal burn.  How would I growing
>> up in California?  Come to think of it, I don't think I've seen
>> coal burn since.
> 
> Dude, you really gotta go buy a barbecue. I even know who you probably
> want to test it on... :-)

Heh.  I do know the difference between charcoal (processed
from wood) and coal (some forms of fossilized plants from under
ground) ;-)  Have you ever seen coal burn Jonathan?  I know it
isn't common in California (though of course charcoal is, I
do have a barbeque), but I thought coal was more common on the
east coast?

--Jed  http://www.webstart.com/jed/