[antlr-interest] Re: RFC: slight change in ! semantics

[Terence Parr]

On Friday, October 5, 2001, at 05:10 AM, Ric Klaren wrote
...
classDefinition ! [AST modifiers]
> 	:	"class" IDENT
> 		// it _might_ have a superclass...
> 		sc:superClassClause
> 		// it might implement some interfaces...
> 		ic:implementsClause
> 		// now parse the body of the class
> 		cb:classBlock
> 		{#classDefinition = #(#[CLASS_DEF,"CLASS_DEF"],
> 							   modifiers,IDENT,sc,ic,cb);}
> 	;
>
> The problem is the use of IDENT in the action code and the use of the !
> operator. Because the ast generation is turned of no declaration for the
> AST for IDENT is made, and I have no nice way of finding out that IDENT 
> is
> used in the action (prescanning the action will be ugly AFAIK). I always
> used the ! as nothing gets generated and so I always label these tokens 
> in
> these cases upon reading the docs it turns out that ANTLR indeed 
> supports
> the above (another of these little inconsistencies).

I'll add a few thoughts here.  I can remember wondering if the unlabeled 
reference to a token from an AST action was a good idea.  Can't remember 
if John Lilley or I came up with the "shortcut" of not requiring a 
label, but I definitely remember being nervous about the decision; 
naturally, the decisions in the early days of Java-based ANTLR were done 
in extreme haste given my time.

I turned off code generation on unlabeled token references since they 
couldn't be used in an action...well, that is until Ric reminds of the 
dreaded IDENT reference from AST code.  FYI, I *rarely* do this.  I 
always use a reference since it's explicit.  Probably why I didn't run 
into this.  OTOH, didn't I do the AST stuff for the Java grammar? ;) 
oops.

> So the options are:
> 1) No optimization and generate 'sloppy' code until a later release.
> 2) Label the IDENT and use the label in cases like this.
> 3) always generate declarations for things like IDENT (not for things 
> like
>    "class" since you can't reference it in actions (can't?) )

This is a shortcut that will work and a decent compiler will remove the 
code for creating the value if it can determine the ctor is side-effect 
free.  Yeah, right. ;)

Since the AST action is special, it's best to track which tokens are 
referenced and update the "generate AST node" boolean for that grammar 
item.  These actions have to be translated anyway, so a simple check 
inside the action recognizer should fix this.  Oh wait, actually, forget 
that.  You'd have to do 2 passes over the rule looking for backward 
references.  Ick.

Option 3 is good for now and optimization folks can simply rewrite to 
not use the rule level ! and use individual ! on the grammar elements.

Ter


 

Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/