Quoting @mrosoiu

the syntax for method parameter annotations is pain and suffering to my eyes :)

Hrm, in Java it's similar:

public class Example {

    public void setConnection(@Inject(name = "db") DBConnection conn)  {
        // ...
    }
}

See http://docs.oracle.com/javase/1.5.0/docs/guide/language/annotations.html

Same goes for C#:

public class Example 
{

    public void SetConnection([Inject(Name = "db")] DBConnection conn)
    {
        // ...
    }
}

See http://msdn.microsoft.com/library/z0w1kczw.aspx

Is the pain in our syntax proposal caused by the /* ... */?

@mikey179 proposed embedding the parameter name in the annotation and the writing it on top of the method declaration line (or was it in the apidoc) - can't remember exactly where. Anyways, this could be an alternative:

<?php
  class Example extends Object {

    #[@$conn: inject(name= 'db')]
    public function setConnection(DBConnection $conn)  {
      // ...
    }
  }
?>

This is also similar to C# syntax for attribute targets (see above MSDN page, about halfway down the page).

@mrosoiu - how does that look to you?

Timm, the "pain and suffering" is definitely a matter of personal taste. Please don't take it as something against your and @mikey179's efforts on this issue. Plus, it was said in a funny context, with awesome llamas and all :)

I understand the reasons behind the need of /*[...]*/ syntax and the fact that this is a language constraint and there is no alternative... but, same as with the PHP official namespacing schema, I just think it looks ugly. When I can't use XP-Framework, I prefer to write old-school / no-namespace PHP, for the simple fact that the backslashes hurt my eyes.

Regarding your alternative proposal with parameter annotations included in method annotations: I've been playing with it a bit and it doesn't look bad at all; not sure about performance / speed impact, but I definitely like it more than /*[...]*/

@mrosoiu - @mikey179 and me discussed this yesterday in the bar, and we also prefer the #[@$conn: inject(name= 'db')] syntax... I'll incorporate this back into the RFC, I think.

XP Language way of doing this @ xp-lang/compiler#4