Subj : Re: Another JS question
To   : netscape.public.mozilla.jseng
From : Issac Goldstand
Date : Sun May 09 2004 01:51 am


"Brendan Eich" <brendan@meer.net> wrote in message
news:409A7DC5.8030109@meer.net...
> Issac Goldstand wrote:
>
> > JSIDArray *jids=JS_Enumerate(cx,obj);
> > for (int i=0;i< jids->length;i++) {
> >     if (JS_IdToValue(cx,jids->vector[i],&rval)) {
> >         ... examine and process rval ...
>
> You'd then call JS_GetUCProperty or JS_GetElement, depending on whether
> JSVAL_IS_STRING(rval) -- if it is, call JS_GetStringChars and
> JS_GetStringLength to gather the name and length parameters to
> JS_GetUCProperty; else JSVAL_IS_INT(rval) and you can index into the
> object using JSVAL_TO_INT(rval).

Does that mean that the array of property IDs returned from JS_Enumerate
will be jsvals which contain either the property name in a JS_String or the
numeric element ID of the element?  Is this always true?  Can there be both
in an object?  (I'd imagine yes, as my own example proves)

>
> >     } else {
> >         ... handle unexpected "holes" in JSIDArray
>
> No, if JS_IdToValue returns false (which can't happen in the current
> codebase, btw), there was an error and you should propagate it.  That
> says to test if (!JS_IdToValue(...)) and return JS_FALSE, avoiding the
> else clause.

I would have thought to use the else to either plug an UNDEFINED value or
spew some (hopefully) somewhat useful message to STDERR...

  Issac

.