Subj : Re: The root of the syntax tree
To   : netscape.public.mozilla.jseng
From : joao
Date : Wed Jul 07 2004 12:02 pm

Hi Brendan,

Brendan Eich wrote:
> Not when you are generating code.  To conserve and recycle JSParseNode 
> space, each top-level statement is compiled in turn: parsed and then 
> traversed by js_EmitTree to generate bytecode.  If you want the entire 
> tree, don't use a JSCodeGenerator and don't call js_CompileTokenStream 
> -- call js_ParseTokenStream instead.

Thanks for the explanation, I obviously jumped onto the first piece of
code I found, I could have looked better.

>> Related question : for a string literal, I get a TOK_STRING token with
>> a JSOP_STRING operator, but I can't see how to get at the value of the
>> string ; it has arity PN_NULLARY so there's nothing more in the
>> JSParseNode. How can I reach this value ?
> 
> pn->pn_atom is the string's atom.  Looks like jsparse.h's big comment 
> does not say this clearly enough.  But if you read very carefully, 
> noting the commas in the first column, you can see it.  Look for 
> "pn_atom: name, string, or object atom" in the third column.

You're right, of course. I see now that the comment groups NAME, STRING,
and OBJECT together, under the 'name' variant. But this is strange :
when I look inside the JSParseNode, token type is TOK_STRING but
pn_arity is PN_NULLARY, not PN_NAME... and yet, there is a
pn_u.name.atom field. I assumed pn_arity would be the discriminating
field for the pn_u union ?

Shouldn't the TOK_STRING have arity PN_NAME ?

> 
> /be

Thanks again,
Joao

-- 
joao at teaser dot fr

.