http://d.puremagic.com/issues/show_bug.cgi?id=2511

           Summary: dmd finds a proper overload as a non-covariant
           Product: D
           Version: 2.021
          Platform: PC
        OS/Version: Windows
            Status: NEW
          Keywords: rejects-valid
          Severity: normal
          Priority: P2
         Component: DMD
        AssignedTo: bugzilla@digitalmars.com
        ReportedBy: 2korden@gmail.com


Here is a sample code that triggers an error:

// File I.d
module I;

import Base;

interface I
{
}

interface SubI : I
{
}

// File Base.d
module Base;

import I;

class Base
{
    I create() {
        return null;
    }
}

class Derived : Base
{
    SubI create() {
        return null;
    }
}

Compilation options: "dmd I.d"
Note that "dmd Base.d" compiles okay.


-- 

http://d.puremagic.com/issues/show_bug.cgi?id=2511





------- Comment #1 from 2korden@gmail.com  2008-12-13 10:22 -------
And here is an error message:

Base.d(15): function Base.Derived.create of type SubI() overrides but is not
covariant with Base.Base.create of type I()


-- 

http://d.puremagic.com/issues/show_bug.cgi?id=2511


2korden@gmail.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|normal                      |blocker




------- Comment #2 from 2korden@gmail.com  2008-12-13 10:32 -------
The error can fixed by removing "import Base;" from the I.d, but unfortunately I can't do this in my application because interfaces have methods that accept and return references to Base and Derived:

interface I
{
    void doStuff(Base b);
}

interface SubI : I
{
    void doStuff(Derived b);
}

A possible solution is to merge them into a single file (which is not acceptable). You should also have a proper file order to compile them at once:

dmd Base I // okay
dmd I Base // fails to compile

Rising its severity until I find a workaround.


-- 

http://d.puremagic.com/issues/show_bug.cgi?id=2511


smjg@iname.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |smjg@iname.com
            Summary|dmd finds a proper overload |Covariant return type
                   |as a non-covariant          |doesn't work with circular
                   |                            |import
            Version|2.021                       |1.043




------- Comment #3 from smjg@iname.com  2009-04-13 16:40 -------
This sounds related to old bug 125.  And it occurs in the D1 line too.

Also rewriting the summary line as the one that was there was confusing.


-- 

http://d.puremagic.com/issues/show_bug.cgi?id=2511



--- Comment #4 from Rainer Schuetze <r.sagitario@gmx.de> 2009-09-18 01:07:31 PDT ---
Created an attachment (id=456)
better detection if a base class is forward referenced

The current implementation (as of DMD 2.032) defers covariance analysis only if the base class is not yet semantically analysed and it is not an interface. The patch refines this condition by testing whether the class has any forward referenced base classes or interfaces.

-- 
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=2511


Walter Bright <bugzilla@digitalmars.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |bugzilla@digitalmars.com


--- Comment #5 from Walter Bright <bugzilla@digitalmars.com> 2010-08-27 22:33:25 PDT ---
The patch breaks this code:

class UA {
    A29 f() { return null; }
}

class UB : UA {
    B29 f() { return null; }
}

class A29
{
}

class B29 : A29
{
}

-- 
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=2511



--- Comment #6 from Walter Bright <bugzilla@digitalmars.com> 2010-08-27 22:49:56 PDT ---
Partial apply of patch:

http://www.dsource.org/projects/dmd/changeset/638

-- 
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=2511



--- Comment #7 from Rainer Schuetze <r.sagitario@gmx.de> 2010-08-28 00:23:08 PDT ---
Oh, it's only been tested on interfaces. ClassDeclaration::isBaseInfoComplete() is broken, class Object doesn't need a base class:

int ClassDeclaration::isBaseInfoComplete()
{
    if (!baseClass)
        return ident == Id::Object;
...

-- 
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=2511



--- Comment #8 from Walter Bright <bugzilla@digitalmars.com> 2010-08-29 12:55:08 PDT ---
This now fails:

class UA {
    A29 f() { return null; }
}

class UB : UA {
    B29 f() { return null; }
}

class A29
{
}

class B29 : A29
{
}

-- 
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=2511


Brad Roberts <braddr@puremagic.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |braddr@puremagic.com


--- Comment #9 from Brad Roberts <braddr@puremagic.com> 2010-08-29 13:08:40 PDT ---
NOTE: these tests are part of the dmd svn depot these days.. so if you check it out from dsource you can run them yourself

cd $(dir}/dmd/src
make
cd ../test
make

It's only known to work on linux, though ought to work on any posix system and will need work to work on windows.  Any fixes that are required I'd be happy to fold in.

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