too many contacts, part II.

[chang@cme.nist.gov (Forrest Chang)]

Gentleman,
	
	It seems further investigation of my "too many contacts" problem
narrows the cause but not the effect...

	I recreated the CLX program for clue, i.e. a program that lets me
generate and map/present contacts.  This one it seems will not let me make
more than 400 contacts, but this one seems to crash the client, and not 
the server.

	The number of contacts it allows me to create varies, apparently to
how fast they are created (I create them with key-presses).  This is slightly
confusing.

	The clx pgm, allows me to generate up to 1485 windows, the side 
effect is slowing down the server (understandably).

	The original program in question (a tree editor in which each node
is a contact) crashed the server when it read in a predefined tree of
some substatial size or several small trees.  However, when I tried to isolate
the problem, by making nodes one at a time, it first locked up the client,
but when I killed it with xkill, the server crashed.

	These problems were not present in the program with clue 1-13, and
appeared when we moved over to clue 1-15.  Has the contact implementation 
changed a fair amount between versions?  The problem was origninal diagnosed
on a Sun 3/60 running X11R3, the recent testing has been on a sun2 running
X11r3.
	Kerry Kimbrough has never seen this problem.  Is there anyone else
out there with a similar problem.  Can anyone else reproduce a crashing of the
client or server with the creation of, say, 400+ contacts. 
	
	Does anyone involved think clue 6.0 will help, or should we either
swithc back to 1-13 or, not use contacts for the nodes?

	Forrest Chang

	NIST