On 02/24/2005 07:26 AM, Jeff Flinn wrote: [snip]
I thoroughly agree with all you've stated. I think this is the best foundation for a generic tree library effort that I've seen. In particular I too would think it imperative that the interaction with existing STL and BGL algorithms be supported. I hope Justin would look at using your concept organization as a starting point.
Could one assume that a tree library implemented as described would be usable with Spirit for AST's and parse trees?
Could it also be used with or merged with multi_array. IOW, a multi_array, m, can be thought of as a tree, where tree depth is m.num_dimensions() and each level of the tree has the same depth and number of children. This is not all that far fetched, as I think it could be used to calculate the follow sets of a Spirit Grammar since the follow set is defined in matrix terms (with transpose and closure operations) by Thomas W. Christopher [ p. 26 of http://facweb.cs.depaul.edu/tchristopher/grmanl.pdf ] Since the grammar would be stored as some sort of tree, which can be thought of as a sparse matrix or sorts, then calculating the transpose and closure would be matrix (or multi-array) operations.