Re: [RFC] Allow "use( $longname as $alias)" syntax for closures declaration

On 2012-04-13, Nicolas Grekas <nicolas.grekas+php@gmail.com> wrote: > > Yes, the original used "$this->getFooBar()" -- but it could have been > > "$someOtherObject->getFooBar()" > > Yep, you and Anthony are right, my comment was stupid... > > Then I have an other objection :) > > I'm not comfortable with mixing declarative lines with expressions: > function (..) use (..) is always purely declarative, whatever the (..) > but > function (..) use ((expression) as $foo) isn't any more. > > I would feel the same e.g. with a syntax like: > function ($a = $foo->bar()) {} (a closure with default argument calculated > once but give as expression.) > > If you like "use ((expr) as $foo)", do you like the above? > This is an open question, a yes would be coherent, > but otherwise, why allow one and not the other? No -- because you can't do it when declaring a normal function (non-closure/lambda/anonymous). If we could do that syntax with regular function/method declarations: // lazy-create $foo if not passed public function doSomething(FooBar $foo = new FooBar()) it _might_ make sense, but as it stands, it's inconsistent with how we allow function declarations. I'm on the fence about "use ((expression) as $foo)" -- I fully like the idea of aliasing closure variables, but still thinking on the expression syntax. It would be _inconsistent_ with how it works with namespaces (which uses literals only), and _semi-consistent_ with traits (which have the "insteadof" syntax). I can definitely think of some nice use cases surrounding them, though. -- Matthew Weier O'Phinney Project Lead | matthew@zend.com Zend Framework | http://framework.zend.com/ PGP key: http://framework.zend.com/zf-matthew-pgp-key.asc