New algorithm for large splits

[johnwhitington]

Say a file of 8000 pages and 200Mb is being split into individual pages. There is too much graph traversal going on. Can we cache? If obj 20 implies objs 21, 22, 23 why can't we cache that?

[johnwhitington]

Or, some method which takes all the possible split points and, in one sweep, calculates the sets of objects. i.e we want the transitive closure of the graph.