http://d.puremagic.com/issues/show_bug.cgi?id=3452

           Summary: Can't alias member functions such that the object name
                    is implicitly stored in the alias
           Product: D
           Version: 2.035
          Platform: Other
        OS/Version: Windows
            Status: NEW
          Severity: normal
          Priority: P2
         Component: DMD
        AssignedTo: nobody@puremagic.com
        ReportedBy: dsimcha@yahoo.com


--- Comment #0 from David Simcha <dsimcha@yahoo.com> 2009-10-29 14:00:43 PDT ---
struct Foo {
    void bar() {}
    void baz() {}
}

void doStuff(Foo foo) {
    alias foo.bar fun;
    fun();  //  Error: need 'this' to access member bar
}

I can't think of any reason why this shouldn't work.  foo.bar is a compile-time
symbol for the member function Foo.bar() on the instance foo.  Once I alias
foo.bar to fun, calling fun() should be equivalent to calling foo.bar().

Even if there's some language legalese reason why this shouldn't work according to the spec, it's still a reasonable enhancement request.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------

http://d.puremagic.com/issues/show_bug.cgi?id=3452


Andrej Mitrovic <andrej.mitrovich@gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |samukha@voliacable.com


--- Comment #1 from Andrej Mitrovic <andrej.mitrovich@gmail.com> 2013-02-08 15:36:42 PST ---
*** Issue 7828 has been marked as a duplicate of this issue. ***

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------

http://d.puremagic.com/issues/show_bug.cgi?id=3452


Dicebot <public@dicebot.lv> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |public@dicebot.lv


--- Comment #2 from Dicebot <public@dicebot.lv> 2013-08-06 09:46:54 PDT ---
Another application is template alias parameter:

string boo(alias T)()
{
    return T.stringof;
}

struct A
{
    int field;
}

pragma(msg, boo!(A.field)());

currently this fails with "Error: need 'this' for 'boo' of type 'nothrow @safe string()'" which does not make sense because compile-time usage of symbol "A.field" does not require "this" pointer.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------

http://d.puremagic.com/issues/show_bug.cgi?id=3452


Maxim Fomin <maxim@maxim-fomin.ru> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |maxim@maxim-fomin.ru


--- Comment #3 from Maxim Fomin <maxim@maxim-fomin.ru> 2013-08-06 10:12:00 PDT ---
(In reply to comment #0)
> struct Foo {
>     void bar() {}
>     void baz() {}
> }
> 
> void doStuff(Foo foo) {
>     alias foo.bar fun;
>     fun();  //  Error: need 'this' to access member bar
> }
> 

Actually error message is reasonable for the same reason as Foo.bar requires this pointer. The root of the issue is that alias does not capture local variable in this context - it captures only type name. The code above is essentially alias Foo.bar fun and since bar() is nonstatic the code doesn't compile.

> I can't think of any reason why this shouldn't work.  foo.bar is a compile-time
> symbol for the member function Foo.bar() on the instance foo.  Once I alias
> foo.bar to fun, calling fun() should be equivalent to calling foo.bar().

Because of underspecification many view things different to what compiler does.

> Even if there's some language legalese reason why this shouldn't work according to the spec, it's still a reasonable enhancement request.

Yes, but this isn't a minor enhancement.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------

http://d.puremagic.com/issues/show_bug.cgi?id=3452



--- Comment #4 from Dicebot <public@dicebot.lv> 2013-08-07 06:27:12 PDT ---
(In reply to comment #3)
> The root of the issue is that alias does not capture local
> variable in this context - it captures only type name. The code above is
> essentially alias Foo.bar fun and since bar() is nonstatic the code doesn't
> compile.

There is a big issue with `alias` specification because of no clear definition what is captured. Current documentation simply describes behavior of reference implementation in various cases which is rather inconsistent on its own (as far as I am aware, there is no even common alias handling in dmd).

Now my understanding of the `alias` concept is simple - it should capture symbols, be it type symbol or variable symbol or anonymous lambda literal symbol.

But I think you are right - cleaning this is worth separate DIP, plenty of corner cases will arise.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------