Re: parent pointers in AST nodes

Torben said:


> I find the visitor pattern non-intuitive for AST processing. A set of
> mutually recursive functions (one for each syntactic category, i.e., one
> for expressions, one for statements etc.) that each do case analysis
> over the node type and calls recursively on children nodes is much more
> intuitive.


For my own work with post-parse trees (AST or otherwise), I find that
a single visitor with registered node-type passengers proves to be the
most efficient. This allows for a simultaneous (for some value of
"simultaneous") top-down, bottom up, breadth-first/last pass, with
each passenger, and reduces overall processing time accordingly. It
does, however, require passengers to be registered in the right order
if visitation order dependencies are important and is especially
opaque if a passenger actually modifies the tree during its visit to a
node.


But overall, "it works here."


Quinn

Return to the comp.compilers page.
Search the comp.compilers archives again.