It's kind of a scope issue. Here is the test java script:

var Util = {
bar: function() {
function it() {
return "I'm in it()";
}
function g() {
println("This is bar! - " + it());
}
g();
}
}
Util.bar();

It runs well on IE & Mozilla but failed on your JS engine which reports
"property [it] is undefined and no call method". I've to add 'this' in front of
it() to make it work. But I think this is not necessary since function it() is
defined in the same scope as function g(). It seems like it() will only look for
global scope if there is no 'this' in front of it.

Can you please expain the details and point me a way to solve this issue since we found many scripts on web will not have 'this' when refer to an inner function.

Thanks!

"adzheng" <adzheng_member@pathlink.com> wrote in message news:dn7umc$2pal$1@digitaldaemon.com...
> It's kind of a scope issue. Here is the test java script:
>
> var Util = {
> bar: function() {
> function it() {
> return "I'm in it()";
> }
> function g() {
> println("This is bar! - " + it());
> }
> g();
> }
> }
> Util.bar();
>
> It runs well on IE & Mozilla but failed on your JS engine which reports "property [it] is undefined and no call method". I've to add 'this' in
front of
> it() to make it work. But I think this is not necessary since function
it() is
> defined in the same scope as function g(). It seems like it() will only
look for
> global scope if there is no 'this' in front of it.
>
> Can you please expain the details and point me a way to solve this issue
since
> we found many scripts on web will not have 'this' when refer to an inner function.

The 'scope chain' is described in ECMA 262 v3 10.1.4: "... the scope chain of the execution context is affected only by with statements and catch clauses." In other words, nested functions do not add to the scope chain.

g() is found because it is a member of the 'activation object' for the
function it is nested in. Inside g(), the it() function is not part of the
activation object created for g().

The script code is apparently relying on bugs in IE and Mozilla.

Thank you, Walter, for your quick reply.

In order to be consistent with other scripts which run well on IE&Mozilla we've to find a way to solve this issue and we did it.

Here are some evidence found on web:

http://www.webreference.com/js/column80/6.html

You can see that in the section on scope, it says:
------------------------------------------------
Function code is code in the body of a function. The scope of function code is not trivial. It includes:

* The variable object of the calling code's execution context. This object
includes all variables that were defined up to the point where the function (or
its outer function) is called.
* The variable object of the outer function, if there is an outer function.
* The global object which is the browser window object.
* The arguments object which includes all passed parameter values.

The 2nd asterisk stands the point.

Anyway, thanks a lot for your help. We probably need your help to solve other problems in the future!

In article <dnb7ah$3pc$1@digitaldaemon.com>, Walter Bright says...
>
>
>"adzheng" <adzheng_member@pathlink.com> wrote in message news:dn7umc$2pal$1@digitaldaemon.com...
>> It's kind of a scope issue. Here is the test java script:
>>
>> var Util = {
>> bar: function() {
>> function it() {
>> return "I'm in it()";
>> }
>> function g() {
>> println("This is bar! - " + it());
>> }
>> g();
>> }
>> }
>> Util.bar();
>>
>> It runs well on IE & Mozilla but failed on your JS engine which reports "property [it] is undefined and no call method". I've to add 'this' in
>front of
>> it() to make it work. But I think this is not necessary since function
>it() is
>> defined in the same scope as function g(). It seems like it() will only
>look for
>> global scope if there is no 'this' in front of it.
>>
>> Can you please expain the details and point me a way to solve this issue
>since
>> we found many scripts on web will not have 'this' when refer to an inner function.
>
>The 'scope chain' is described in ECMA 262 v3 10.1.4: "... the scope chain of the execution context is affected only by with statements and catch clauses." In other words, nested functions do not add to the scope chain.
>
>g() is found because it is a member of the 'activation object' for the
>function it is nested in. Inside g(), the it() function is not part of the
>activation object created for g().
>
>The script code is apparently relying on bugs in IE and Mozilla.
>
>

"adzheng" <adzheng_member@pathlink.com> wrote in message news:dnckk9$2tp8$1@digitaldaemon.com...
> Here are some evidence found on web:
>
> http://www.webreference.com/js/column80/6.html

The web site seems to be down at the moment.

> You can see that in the section on scope, it says:
> ------------------------------------------------
> Function code is code in the body of a function. The scope of function
code is
> not trivial. It includes:
>
> * The variable object of the calling code's execution context. This object includes all variables that were defined up to the point where the
function (or
> its outer function) is called.
> * The variable object of the outer function, if there is an outer
function.
> * The global object which is the browser window object.
> * The arguments object which includes all passed parameter values.
>
> The 2nd asterisk stands the point.

I can't find support for (2) in the ECMA 262 v3 standard.

"adzheng" <adzheng_member@pathlink.com> wrote in message news:dnckk9$2tp8$1@digitaldaemon.com...
> In order to be consistent with other scripts which run well on IE&Mozilla
we've
> to find a way to solve this issue and we did it.

What was the solution you found?

In fact we modified the scope chain.

In article <dncn6k$30e$2@digitaldaemon.com>, Walter Bright says...
>
>
>"adzheng" <adzheng_member@pathlink.com> wrote in message news:dnckk9$2tp8$1@digitaldaemon.com...
>> In order to be consistent with other scripts which run well on IE&Mozilla
>we've
>> to find a way to solve this issue and we did it.
>
>What was the solution you found?
>
>

Can you post the diffs?

"adzheng" <adzheng_member@pathlink.com> wrote in message news:dnvnde$2vpi$1@digitaldaemon.com...
> In fact we modified the scope chain.
>
> In article <dncn6k$30e$2@digitaldaemon.com>, Walter Bright says...
>>
>>
>>"adzheng" <adzheng_member@pathlink.com> wrote in message news:dnckk9$2tp8$1@digitaldaemon.com...
>>> In order to be consistent with other scripts which run well on IE&Mozilla
>>we've
>>> to find a way to solve this issue and we did it.
>>
>>What was the solution you found?
>>
>>
>
> 

Here is the code snip. Please let me know your opinion. Thanks! So far we haven't found any side effects.

unsigned scoperootsave;
case IRcallscope:     // a = s(argc, argv)
...
scoperootsave = cc->scoperoot;
cc->scoperoot = scope->dim;
...
cc->scoperoot = scoperootsave;

In article <do74c9$1gj9$1@digitaldaemon.com>, Walter Bright says...
>
>Can you post the diffs?
>
>"adzheng" <adzheng_member@pathlink.com> wrote in message news:dnvnde$2vpi$1@digitaldaemon.com...
>> In fact we modified the scope chain.
>>
>> In article <dncn6k$30e$2@digitaldaemon.com>, Walter Bright says...
>>>
>>>
>>>"adzheng" <adzheng_member@pathlink.com> wrote in message news:dnckk9$2tp8$1@digitaldaemon.com...
>>>> In order to be consistent with other scripts which run well on IE&Mozilla
>>>we've
>>>> to find a way to solve this issue and we did it.
>>>
>>>What was the solution you found?
>>>
>>>
>>
>> 
>
>