On Jul 23, 4:06 am, Dave Methvin <[EMAIL PROTECTED]> wrote:
> All of those results make sense.
>
> > "print( (new String('foo') === new String('foo')))"
> > false
>
> Those are two different objects, even though they have the same value;
> object1 !== object2 by definition of the === operator.
Agreed.
> In real code, I have never seen a good reason to use 'new String()'
> rather than a primitive string. There are are several situations where
> 'new String()' will break code that seems like it should work
> perfectly well--for example, in an eval() or switch statement.
Agreed as well, but new String() does turn up unexpectedly from time
to time. Try this code with jQuery 1.1.3.1:
alert( $(mySelector).css( 'background-color', null ) )
That misbehaves and acts like a getter, and returns the color using
(new String()). (i had to spend some time debuggering that problem in
an application.)
> "print('foo' === 'foo')"
> true
>
> Two primitives with the same type and value are === as well as ==
i probably would expect it for string constants, but i would not
expect it to be true for computed strings:
[EMAIL PROTECTED]:~$ SpiderApe -e "a=function(){return 'abc'[0];};
print(a() === 'a')"
true
> > "print(typeof someUndefined == 'fuzzyDuck')"
> > false
>
> typeof someUndefined is 'undefined', and 'undefined' != 'fuzzyDuck'
This makes sense, of course, but seems to contradict the standard (as
Rob described it above).
> > "print(typeof someUndefined == 'undefined')"
> > true
>
> typeof someUndefined is 'undefined' (string) so the two are equal.
This is a case where i would not expected === to work because typeof
computes a value and 'undefined' is a constant string literal that the
JS compiler can intern/pool (whereas the 'undefined' returned from
typeof is a native C string somewhere in the engine).
> > "print(typeof someUndefined == undefined)"
> > false
>
> The string 'undefined' is never equal to the intrinsic undefined
> value.
i find it unfortunate that typeof returns a string 'undefined' for the
undefined case, similarly to how the sqlite3 C library has a function
called sqlite3_error() which returns the last error string from the
library and evaluates to the string "not an error" when there is no
error (instead of using a null string, as would have been more
sensible).
Of course, no matter how unfortunate i find it, it's written in stone
in the ECMA standard, so bitching won't do much about it.
:)
