Subj : Re: JS_AddNamedRoot
To   : netscape.public.mozilla.jseng
From : Brendan Eich <brendan@meer.net>
Date : Sun Jan 04 2004 11:41 pm

aadamov wrote:

>Followinf up on the previous message. I did create a generic wrapper for DOM
>embedded in SpiderMonkey, but... Couple of buts:
>1) I would still like to know if there is well thought through "standart"
>implementation for that;
>2) I am stuck trying to use JS_AddNamedRoot in order to instantiate root of
>the DOM object in JS context. In fact I run into problem using
>JS_AddNamedRoot even for standart types. For instance:
>   std::string name = "red";
>   JSString *jstr = JS_NewStringCopyN(cx, name.c_str(), name.length());
>   JS_AddNamedRoot( cx, (void*)jstr, "myRoot" );
>  
>

Pass &jstr, not jstr -- roots are pointers registered by their address.

/be

>   std::string script = "myRoot;";
>   jsval rval;
>   uintN lineno = 0;
>   JSBool ok = JS_EvaluateScript(cx, globalObj, script.c_str(),
>                                  script.length(), "script", lineno, &rval);
>
>Evaluating above script I get JS_FALSE return value.
>I guess it only makes sense, since I am passing a void pointer to jstr in my
>call to JS_AddNamedRoot. How in life would runtime know what type of an
>object I just passed to it. That is unles there is some kind of reflection
>mechanism build into it. Better off going to sleep now.
>
>
>  
>

.