6b0
Subj : Re: What does the MAP_CHECK_SLOT assert mean ?
To   : netscape.public.mozilla.jseng
From : nick.longinow@brio.com (Nick Longinow)
Date : Thu Jan 30 2003 07:22 am

Some more info.  It turns out that the culprit is my native
method for doing OBJ_LOOKUP_PROPERTY, ::LookupProperty, operating
on a native object, which is doing the following (which results in 
a bad 'slot' value for the associated JSScopeProperty value):


    JSProperty value = ...(get from JSObject passed in)
    value.id = id;  /* id is passed in from the js_Interpret method */
                    /* id has the value 33867336 */
    JSScopeProperty *sprop = (JSScopeProperty*)value;
    jsint slot = sprop->slot;   /* slot will be huge - 33844916U  */


So, the long and short of it is that when we return to js_Interpret
with this sprop, it is so huge a slot value that it will trip the
assert in MAP_CHECK_SLOT, which is expecting a slot value much much
smaller (on my Windows build its more in the range of 4U or 6U).  

Question - this is a Solaris 2.8 build of the newer JS 1.5 which I am
trying to now use instead of my older JS 1.4.2 build.  Do I need to do
something different in finding/setting these property values for native
objects ? ie, like initializing something ? 


nick.longinow@brio.com (Nick Longinow) wrote in message news:<83377f55.0301291144.1d4fb64f@posting.google.com>...
> I just ported our code to Solaris 5.8 and am using
> the new JS 1.5.  Any ideas why the following call
> is asserting (and aborting) ?
> 
> #define MAP_CHECK_SLOT(map,slot) \
>     JS_ASSERT((uint32)slot < JS_MIN((map)->freeslot, (map)->nslots))
> 
> The call is coming inside js_Interpret.

.

0