Assigning proc to event at run time

[BubikolRamios]

Code: Pascal  [Select][+][-]

1. procedure Twirl(Sender: TObject);    
2.  
3. EditRadius.OnChange:= @Twirl;    
4.  

This works only if both twirl and OnChange have same parameters.

Can this assignment be done without that condition (both having same parameters) ?

Tested, at design time I can put Twirl at onChange event, regardles if it has parameter/s or not.
I think it complains a bit, but accepts anyway and works.

[Fibonacci]

Code: Pascal  [Select][+][-]

1. {$modeswitch anonymousfunctions}
2. {$modeswitch functionreferences}
3.  
4. // ...
5.  
6. EditRadius.OnChange := procedure(Sender: TObject)
7. begin
8.   // do whatever you want, call any procedure you want
9.   Twirl;
10. end;

[alpine]

Shouldn't it be procedure(...) of object?

[Thaddy]

Yes, of object.
An event expects a hidden self parameter and that is what of object provides.
(You do not have to actually use it, but it must be there.)

[PascalDragon]

Code: Pascal  [Select][+][-]

1. procedure Twirl(Sender: TObject);    
2.  
3. EditRadius.OnChange:= @Twirl;    
4.  

This works only if both twirl and OnChange have same parameters.

Can this assignment be done without that condition (both having same parameters) ?

The compiler has strong typing for a reason. A method/function passed to a method/function pointer with a different signature will lead to crashes depending on the parameters and the calling convention / platform involved.

Tested, at design time I can put Twirl at onChange event, regardles if it has parameter/s or not.
I think it complains a bit, but accepts anyway and works.

It's simply that the compiler can't check at compile time whether everything is correct in the LFM file, because it knows nothing about LFM files. This will break depending on the parameters and the calling convention / platform involved.

Don't do this.

[Bart]

It's simply that the compiler can't check at compile time whether everything is correct in the LFM file, because it knows nothing about LFM files. This will break depending on the parameters and the calling convention / platform involved.

Don't do this.

Indeed don't.
There is a bugreport about Lazarus allowing this/not warning on loading the lfm.
It caused a crash in the reporter's application.

Bart