On 24/01/13 18:45, Jonathan Wakely wrote:
On 24 January 2013 16:21, Alec Teal wrote:
That's because this has nothing to do with objects, in the paper that was
linked (called "strong typing") it implemented new types rather like
objects, "using score = public int { //definitions }; for example,
"extending an int" effectively, this is what I mean by a PHP-noob going
class-mad.
I think you've misunderstood the proposal. The syntax you refer to
defines a strong typedef for int and defines an overloaded operator
for it, it doesn't go class-mad and has nothing to do with PHP or
"everything is a class". It would add no overhead over an int either,
if implemented sensibly, it would only have compile-time properties
and not alter runtime behaviour. Reel in the hyperbole.
....What? No.
In PHP when people learn about objects they tend to make EVERYTHING into
an object and never use inheritance. I'm playing on that! In the paper
they define X Y and Z components, EVERYTHING is a defined type of sorts.
When it comes to runtime overhead, yes there is none in both cases, but
it's about look, the one in the document looks like a class, it makes
sense to be but it's a bulky syntax for something as small as a typedef.
I'm not saying "not" the above, just that it invites overuse - I do have
an extension to the idea that employs more of the abstract algebra
that'd look a lot like what that paper is on about, anyway, I propose
something simple, not different, small and not able to be over-used
without real stupidity taking over (probably)
See what I mean? Again not saying 1 or the other, just that with this,
then the strong ones.
