In this paper Miller Puckette identifies a problem in the relationship between "data" and "process".  In doing so, he associates data with composition and process with performance. 


The problem seems to be how to coordinate or generalize access to data structures and lists of data structures.  A simple iterate and increment seem to work fine, but a search for an element and change it  type operation requires a lot of gymnastics.


Miller, in the end, talks about pointers being a way to solve the problem.  Not sure what the point is, except to note that there is a problem with a simplified approach to access and modification of data lists and data structures within the dataflow paradigm.


Another possible solution to this problem is to incorporate a Mathlab style looping mechanism.  The key idea would be to separate the indexing from the data manipulation.  A special iteration sub patch could be provided that takes as arguments the loop start, end test, and increment.  This type of sub patch would be similar to a C language for loop.  The internals to the patch would be the code that operates on each individual element that the iterating variable visits.  Relations between other values in the list could be accessed through the addition of other iterating variables through the mechanism of nesting of sub patches or through additional variables added to the sub patch (in the same manner as is accomplished in the C language).