[photran-dev] Parser and symbol tables; status of refactoring support

[Jeffrey Overbey <subscribe-photran-dev@xxxxxxxxxxxx>]

FYI if you grab the latest Photran from HEAD, the parser speed has 
increased by a factor of more than 1000, and the symbol table code has 
been moved into the refactoring plug-in and Java 5'ed.  You now need to 
use a RefactoringFortranProcessor to create a parse tree or symbol 
table; the FortranProcessor only does parsing for the CDT model (which 
is now very fast since it builds the model without building an entire 
parse tree first).  The SymbolTableVisitor class is now type-parametric. 
 Also, my parser generator now has support for error recovery, so we 
should be able to tolerate vendor-specific extensions (e.g., %val and 
%ref) by augmenting the grammar with error productions.  We should 
(theoretically) even be able to refactor around them, if we warn the 
user that we don't know what to do with the "erroneous" code.

We are working towards a solid Rename refactoring, which means the 
symbol table code (which is currently very incomplete) will be becoming 
more robust over the next several weeks.  Our next refactoring will be 
Introduce Implicit None, which will also require us to finish the 
reindentation code.

Jeff