Circular Multi-Orgls & Logic Programming

[Mark S. Miller <mark>]

Date: Sun, 22 Oct 89 11:09:04 PDT
   From: heh (Hugh Hoover)

   My feeling is that we won't be needing circular structures (in orgls) very
   soon, and they can be emulated (if necessary) with links.  

I agree completely.  Circularity would be neat, but is certainly not
badly needed.

   Any FE that used
   such a structure would have to go out of its way, but not too far.
     Accepting circularity in the long term may be reasonable.  It is certainly
   better than n^2 checks.

Hmmm.  I have a feeling that the reason we'll be able to do a lot
better that the logic programming guys is that in many ways "inform"
is a lot less ambitious an operation than "unify".  More on this after
Dean & I reconstruct the algorithm we thought of.

     After thinking for a few minutes, I can't come up with a (reasonable)
   scenario in which a partial orgl would become informed to produce a circular
   structure.  Of course, I expect most informs to be with IDs rather than orgls,
   so that eliminates (direct) circularity.  Unless I'm missing something, I don't
   see cases in our initial docs and links that CAN produce a circularity (with
   the caveat that I understand that our current use of recorders is to inform
   them with ID's (stamps or berts) rather than orgls).

No, I also cannot think of a scenario in which the docs&links layer
could produce such a circularity.  However, I would *hate* for the
integrity of the orgls&berts layer to depend on the behavior of the
docs&links layer.