1. specifications to allow user to jump to a place in tree, other than
	the node when the program starts. this means data will not be lost
	before the tree syncs with the data.

2. same as one, but not at the beginning of the program, e.i. allow jumps
	later in the program.  the utility of this feature is not known]
	at this time, but should be considered.

3. the tree program will read an process its input in an awk like manner.
	the user must specify the field at which the key string resides.
	a very slight modification to glnstr will accomplish this. i
	have the modification in mind (dz).

4. a "wait" facility which steve feels would be built into the nodes.
	this would allow the tree pointer to remain at a node thru
	a specified number of data lines, trying to move to children
	while waiting. after the wait period was expired, it would 
	move anywhere (i.e. by the rules).

5. the cumulated weighted pathlen of each node should be optionally/always
	printed at completion of the job. e.g.
		rstr 	times traversed(i.e. matched key)
		___	_______________
		rstr1	    45
		rstr2	     8
		rstr3	   193
		etc
	actually this may not be a good idea since it would be a trivial
	user program to count return strs and tally them????
