Neuer FixInsight ist da!

[Stevie]

Zitat von Der schöne Günther:

Generic interfaces cannot have a GUID.

They can - their use is just limited because all instantiations have the same GUID - which can be dangerous but also useful when you know what you are doing.
However Roman might want to disable triggering that warning for generic interfaces.

[Whookie]

Nice idea, gave it a short try and here are my first expiriences:

W522 does enforce "override" to be written in lower case (which is not required)
W521 gives an error with:

Delphi-Quellcode:

function test: boolean;
var
  LResult: Boolean;
begin
  try
    LResult := TRUE;
  finally
    Result := LResult;
  end;
end;

W504 does warn on missing inherited in a constructor directly derived from TObject (even though it might be good practice(?)) I never call them in that circumstance.

[Der schöne Günther]

Zitat von Whookie:

even though it might be good practice(?) I never call them in that circumstance.

I always explicitly derive from TObject and call the (empty) inherited constructor. Java, for example, is different: It implicitly calls the inherited parameterless constructor (if there's any). Otherwise, it's a compile time error.

Another reason is that it often happened to me that I omitted the (then redundant) call to the super constructor but later decided to derive the class from something other than TObject . Then, it's easy to miss adding the call to the inherited constructor back in. (FixInsight to the rescue! )

[Whookie]

Zitat von Der schöne Günther:

...but later decided to derive the class from something other than TObject . Then, it's easy to miss adding the call to the inherited constructor back in. (FixInsight to the rescue! )

That's one hell of an argument...