On 2009.07.07., at 22:37, Greg Brown wrote:
js> new String("a") == new String("a")
false
js> new Number(2) == new Number(2)
false
...
Which is due to the fact that if any of the operands is a JS
primitive type (i.e. string, number) then the other will be coerced
into a primitive value and then the equality among primitive values
will be used instead.
Yeah, it's bloody un-intuitive, but that's the way JS works.
Nothing you can do about it.
That makes sense. But what if one or both of the operands were Java
objects? e.g. replace the JS native String above with
java.lang.String:
js> new java.lang.String("a") == new java.lang.String("a")
If this expression evaluated to true, it wouldn't violate JS
equality semantics, since JS values are not involved.
Except they are. These values belong to the JavaScript "object" type:
js> typeof(new java.lang.String("a"))
object
They *have* to be exposed as such, since in JavaScript, every value
must belong to one of the JS types, and the *only* types in JS are
these: Undefined, Null, Boolean, Number, String, and Object [1].
Therefore, the == operator will compare two JS objects, for which the
only valid comparison is the reference comparison.
Groovy invokes the corresponding Java method on the left-hand
operand. So, if a Java object was on the left, equals() could be
called, and if a JS object was on the left, then JS reference
equality could be used...
Well, Groovy is a language designed from the start for tight
cooperation with Java. JavaScript - despite the name, is not. Not at
all. For long time, most of its implementations were written in C.
We're doing what we can with Rhino, but there's a good deal of
language semantics mismatch that we just can't hide.
Attila.
[1] Actually, there's also three other value types: Reference, List,
and Completion, but they can not ever be observed by a JS program
itself, they're internal to the runtime. Actually, they need not even
be implemented proper by a JS runtime, but a conformant JS runtime
must behave in certain calculations as if it operated on them. They're
primarily a specification device.
--
twitter: http://twitter.com/szegedi
weblog: http://constc.blogspot.com
home: http://www.szegedi.org
_______________________________________________
dev-tech-js-engine-rhino mailing list
[email protected]
https://lists.mozilla.org/listinfo/dev-tech-js-engine-rhino
