function overloading feature request
lazarus / 64-FPC development

----------------------------------------------------------------------
ad1mt · 2023-12-31 12:56

I've got a function overloading feature request.
I need the compiler to be able to resolve overloaded functions based
on the return type, in addition to the parameter types. So that the
following functions would be syntactically valid, and be resolved
correctly:

Code: Text  [Select][+][-]window.onload = function(){var x1 =
document.getElementById("main_content_section"); if (x1) { var x =
document.getElementsByClassName("geshi");for (var i = 0; i var R:real;
I:integer;function F(p1:integer):integer;overload;  begin {version 1}
end;function F(p1:integer):real;overload;  begin {version 2}
end;beginR:=F(99);  //  <--- this calls version 1 I:=F(99);  //  <---
this calls version 2I'm creating on a library that provides a new data
type, and this feature would be very useful.
Would this be possible?

----------------------------------------------------------------------
wp · 2023-12-31 13:07

Quote from: ad1mt on December 31, 2023, 01:56:27 pmWould this be
possible?
I'm afraid: no. Because a function can be called like a procedure
without assigning its return value to a variable. And if you call the
integer version of the function you can assign its result to a
floating point variable due to internal conversion by the compiler. In
both cases, there is no way for the compiler to distinguish which
overload should be used.

----------------------------------------------------------------------
Josh · 2023-12-31 13:27

my 2cents..

if you had a complex computational line where your result is a real
but your using a mix of real/integer versionof the functions in the
calculation it would either not compile or give spurious result that
could take an age to track down unless you were to cast the function
to use the correct one;

it would be simpler to have the p1 variable set to the datatype,much
easier to debug, and will generate appropriate compiler error if you
have something wrong.

not sure what it would doto existing code.

Code: Pascal  [Select][+][-]window.onload = function(){var x1 =
document.getElementById("main_content_section"); if (x1) { var x =
document.getElementsByClassName("geshi");for (var i = 0; i function
F(p1:longint):longint;overload;begin {version 1}  writeln('v1'); 
result:=p1 div 2;end; function F(p1:Double):Double;overload;begin
{version 2}  writeln('v2');  result:=p1/2.0;end; begin 
writeln(f(99));  writeln(f(99.0):2:2);  readln;end.                 
ps Happy New Year

----------------------------------------------------------------------
MarkMLl · 2023-12-31 13:45

See https://forum.lazarus.freepascal.org/index.php/topic,60218.msg4505
10.html#msg450510 from one of the core developers.

There's more automatic conversion and less sensitivity to the lvalue
type than I for one am happy with: even if the developers agreed we're
pretty much stuck with it.

MarkMLl

----------------------------------------------------------------------
Paolo · 2023-12-31 14:09

but the proposed code compiles

Code: Pascal  [Select][+][-]window.onload = function(){var x1 =
document.getElementById("main_content_section"); if (x1) { var x =
document.getElementsByClassName("geshi");for (var i = 0; i var R:real;
I:integer;function F(p1:integer):integer;overload;  begin {version 1}
end;function F(p1:integer):real;overload;  begin {version 2}
end;beginR:=F(99);  //  <--- this calls version 1I:=F(99);  //  <---
this calls version 1 tooend; 
while this not, and just swapped the order of  listed function

Code: Pascal  [Select][+][-]window.onload = function(){var x1 =
document.getElementById("main_content_section"); if (x1) { var x =
document.getElementsByClassName("geshi");for (var i = 0; i var R:real;
I:integer;function F(p1:integer):real;overload;  begin {version 2}
end;function F(p1:integer):integer;overload;  begin {version 1}
end;beginR:=F(99);  //  <--- this calls version 2I:=F(99);  //  <---
this calls version 2 too and as consequence cannot be compiledend; 
no compiler hint/warning for such situation ? just kept the first
declaration encontered in the code ?

----------------------------------------------------------------------
Leledumbo · 2024-01-01 09:29

Quote from: Paolo on December 31, 2023, 03:09:38 pmbut the proposed
code compiles
Not anymore in the latest trunk, that was an obvious compiler bug.
Both versions are now rejected.

----------------------------------------------------------------------
Paolo · 2024-01-01 12:05

Ok

----------------------------------------------------------------------
PascalDragon · 2024-01-02 22:44

Quote from: ad1mt on December 31, 2023, 01:56:27 pmI'm creating on a
library that provides a new data type, and this feature would be very
useful.
Would this be possible?

No, this would not be possible.

However if you're working on a custom type anyway then you can declare
assignment operator overloads which are essentially the only exception
where this is allowed (cause otherwise they wouldn't make sense).

----------------------------------------------------------------------
Kays · 2024-01-03 12:45

Quote from: ad1mt on December 31, 2023, 01:56:27 pm[…] able to resolve
overloaded functions based on the return type, in addition to the
parameter types. […]If it’s just unary functions:Quote from: Kays on
October 19, 2023, 08:05:34 amQuote from: Kraig on October 19, 2023,
07:31:06 amI want to use same name for functions that return different
types […] Is this possible? […]No, this is not possible; except for
operator overloading you can “select” the := function by its result
data type.Code: Pascal  [Select][+][-]window.onload = function(){var
x1 = document.getElementById("main_content_section"); if (x1) { var x
= document.getElementsByClassName("geshi");for (var i = 0; i {$mode
objFPC}program overlord(input, output, stdErr);        type          
    dummy = record end;                operator := (x: dummy):
string;                begin                        result := 'Foo'; 
             end;                operator := (x: dummy): integer;    
          begin                        result := 42;               
end;                { === MAIN ==================== }        var     
         tmp: dummy;                str: string;                int:
integer;        begin                str := tmp; { assigns 'Foo' }   
           int := tmp; { assigns 42 }        end.

----------------------------------------------------------------------
jcmontherock · 2024-01-03 21:55

Did you try the "out" parameter in the 2 functions definition to be
overriding ?