[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Implementing identification on SableCC ASTs



 --- Jens Kristian Søgaard <jks@cs.auc.dk> wrote: > Hi
Fidel,
> 
> > Well, if that is the case, then this is how you
> solve
> > your problem
> 
> Well, that will really not solve the problem. It
> simply moves the
> problem around (and also make the problem harder).
> 
> To make your solution work, it would be necessary to
> define a
> SymbolTable class that "remembers" everything. I.e.
> when the so called
> SemanticAnalyser class goes through the AST and
> opens scopes, enters
> identifiers and closes scopes - everything has to be
> stored in the
> SymbolTable.
> 
> Normally, the symbol table "forgets" about
> identifiers declared in a
> scope when the scope is closed.
> 
> Making a SymbolTable that "remembers" is trivial
> when using a program
> with a flat block structure. It's more complicated
> with a programming
> language with nested block structure.
> 
> Also you would need for the CodeGenerator to know
> where to open and
> close scopes for identifiers. I.e. this destroys the
> separation of
> concerns.
> 
> It's much simpler to use the HashMap technique
> already employed in
> AnalysisAdapter, and simply pass that around. I.e.
> the solution I
> mention last in my email. I'm using that, and it
> works -- but it doesn't
> seem completely "elegant" to me.


You are totally right. I had forgotten about the local
scope variables as well as nested ones. I haven't
tried the HashMap technique, so I can't really comment
on that.

All the best

Fidel.



__________________________________________________
Yahoo! Plus
For a better Internet experience
http://www.yahoo.co.uk/btoffer