[cap-talk] gauntlet - one way IPC considered practically useless

[Jed at Webstart]

At 01:48 PM 8/27/2006, Norman Hardy wrote:
>I just noticed this 6 months old thread. I was not following the list
>closely at that time.
>I have not entirely digested the content of the thread.
>A similar question arose in another context this April.
>At that time I wrote a note on simplex communications channels into a
>confined compartment.
>See <http://cap-lore.com/CapTheory/KK/Simplex/>.
>The solutions are in the Keykos context. It wasn't hard but neither
>was it trivial.
>The solution relies on the non-standard Keykos subroutine mechanism
>which is a style of 'continuations'.
>(See <http://en.wikipedia.org/wiki/Continuation_passing_style>.)

I started the "gauntlet" thread with this message:

http://www.eros-os.org/pipermail/cap-talk/2006-January/004733.html

I was frustrated with the notion of "do not copy" being considered useful
at all.  I believe this discussion also showed up later in the discussion
about "communicating conspirators":

http://www.erights.org/elib/capability/conspire.html

(which I see I apparently and sadly flipped to "conspiring communicators"
at one point:

http://www.eros-os.org/pipermail/cap-talk/2006-July/005437.html

).

I believe it's been generally accepted that blocking delegation between
bi-directionally communicating conspirators is not possible (dissent?).

However, the gauntlet thread lead into a discussion of ways in which
one-way communication might be useful, particularly in the context of
MLS systems.

I tried to act somewhat as a moderator with this message refocusing
on the above topics with:

http://www.eros-os.org/pipermail/cap-talk/2006-January/004765.html

where I later pick out what I consider the most relevant aspects of
that discussion in:

http://www.eros-os.org/pipermail/cap-talk/2006-January/004839.html

(naturally others may have differing views).  This discussion led to
what I considered an interesting discussion of the "coordination
problem" AKA the "Two Generals Problem":

http://en.wikipedia.org/wiki/Two_Generals'_Problem

where I was able to add some clarity to the historical thread as
noted on the above Wikipedia page (not self serving in this case ;-).

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