call for thoughts: cyclic structures

[Rick Mascitti <acad!blitzen!rick>]

(Please forgive my introduction of new terms for stuff you probably
already have good ones for.  Educate me, please.)

As I stated when giving the netwalk demo last Tue, I'm struggling
with visual representations of cyclicity (in that case it was in
representing the "came from" section of the Topology display).  This
has cropped up in two new areas of thought for me, and I know there
are direct analogues in the InfoFactory, so you folks have probably
wrestled with this already.

Enough babble, here we go....

Situation 1 (analogous to InfoFactory Inclusion List Link Views) is
what I call "Reference Composition", i.e., "show me the links out
from here".  This works just fine if you assume 1.) no cyclicity in
the graph (HA!), or 2.) just show the next (N) level(s).  I can think 
of no really good way to present cyclic connection -- in cflow(1) I
"solved" it by "pointing" to the first reference of any subsequently
referenced entry and deem this barely sufficient for that purpose
(a flow graph generator) but not good enough for a general hypermedia
product.

Situation 2 (analogous to InfoFactory Inclusion List Flat View) is
what I call the "Subsumption Operation", i.e., "give me a view that
subsumes all of the links from here".  This is even worse than the
previous mess (since previously we were only dealing with referents
and here we're trying for containment).  Again either of the two
assumptions work but probably aren't valid.  markm's attempt to join
the two views into a single Inclusion List "squishy" display mitigates
this problem (since you're always only presenting the next level),
but doesn't really solve it -- suppose I want to print (yes, print,
for whatever masochistically baroque reason I might have) out the
document?  (I don't know that this one is solvable without backing
off to a reference.)

Any comments?
rick