Subj : Re: C embedding question: failure in JS_DefineFunction
To   : Pavel Podgoretsky <ppodgoretsky@naturalmessaging.com>
From : Brendan Eich <brendan@meer.net>
Date : Tue Mar 18 2003 11:03 am

Pavel Podgoretsky wrote:

>Hi Brendan,
>
>Sorry I wasn't specific enough.  FindConstructor returns null, which is
>propagated up the stack and eventually JS_DefineFunction returns null.
>There are no other errors.
>

As you say below, that's because you generated a silent failure -- a 
false or null return without a prior JS_Report*Error* or 
JS_SetPendingException call.  Just being clear here (please correct me 
if I've misread you).

>BUT while I was writing up a few more details for you, I found the problem!
>Thank you! :)
>
>The JSClass structure that I use needs a getObjectOps member (countrary to
>what the api docs say!), and that member has to return a non-null
>JSObjectOps.  [There is an access violation otherwise.]
>

That shouldn't be the case.  As you can see from all the JSClass static 
initializers in the engine (date_class, js_ObjectClass, etc.), 
getObjectOps may be null.  Can you say more about *why* you seem to need 
a non-null JSClass.getObjectOps initializer?  Where exactly (file and 
line, and with what stack backtrace) do you crash when you use a null 
initializer?

/be

>  Since I didn't know
>what some of those should do, I just had them return JS_FALSE.  At some
>point, lookupProperty was called [looking for 'Function', for some strange
>to me reason] and returned JS_FALSE instead of setting its outparams to null
>and returning JS_TRUE.  (If I recall correctly, the very helpful comments in
>jspubtd.h were not as helpful when I wrote that code originally...)
>
>Thank you for your extremely prompt response.
>
>-Pavel
>
>
>  
>

.