Subj : Re: [Rhino] can you add default getters/setters?
To   : Mike C.
From : Igor Bukanov
Date : Sat May 08 2004 11:35 am

Mike C. wrote:
> "Igor Bukanov" wrote
> 
>>Well, if "foo(bar) = something" happens is MS-specific scripts often, it
>>makes sense to add it to Rhino to improve support for JScript. It would
>>on the same scale as the implemented support in Rhino for JScript syntax
>>sugar like
>>
>>function object.method() { ... }
>>which is an alias for
>>object.method = function () { ... }
>>
> 
> This syntax works too in JScript:
>       function object.prototype.method() {...}
> so you might want to add it if you touch that area.

Rhino supports that as well. In fact, instead of function name you can 
use arbitrary lvalue.

> 
> 
>>It would also cover an obscure deviation in Rhino from ECMAScript
>>standard as "foo(bar) = something" should throw ReferenceError during
>>script evaluation, not SyntaxError during compilation phase as it
>>happens now.
> 
> 
> As Brendan noted, SpiderMonkey supports this as lhs for native functions, I
> guess it might be usefull if you can add the support in Rhino.
> 
> It does happen often in MS-specific scripts, take the Session example. If
> one wants to share the same code base between JScript and Rhino, it is not
> possible since there are no setter methods for Value in Session, only
> propput or propputref, that cannot be called directly from scripting
> languages. So under JScript I can use only
>       Session("foo") = "bar"
> or manually expand the default setter, as
>       Session.Value("foo") = "bar"
> 
> If I use
>       Session.Value("foo", "bar")
> I get runtime error - Wrong number of arguments or invalid property
> assignment. Note that in the IDL, the Value is declared as
> 
>         [id(00000000), propget]
>         VARIANT Value([in] BSTR bstrValue);
> 
>         [id(00000000), propput]
>         void Value([in] BSTR bstrValue, [in] VARIANT rhs);
> 
>         [id(00000000), propputref]
>         void Value([in] BSTR bstrValue, [in] VARIANT rhs);
> 
> 
> Unfortunately I don't think I can help since Rhino code is quite complex,
> and it might take a while for me to get used to its philosophy. I can just
> pinpoint where the parsing stops, and it is in
> IRFactory.createAssignment(..).
>       int nodeType = left.getType();
> returns Token.CALL, that goes directly to default in the switch, and reports
> the error.

Ok, I suggest to file a bug for this. And note that createAssignment is 
not the only place to patch since I assume Session("foo") += "bar" and 
Session("foo")-- also works in JScript, right?

Regards, Igor

.