[fpc-devel] [Patch/RFC] Warnings for (in/over)complete case statements

[Ondrej Pokorny]

On 12.05.2019 18:56, Jonas Maebe wrote:
> As far as range checking and undefined behaviour is concerned, they 
> do. I.e., you won't get undefined behaviour by assigning any value 
> between low(enum)..high(enum) to them.

Very good, thank you. IMO this should be clearly documented to avoid 
confusion.

And if they are valid they should not cause any runtime errors:

program Project1;
{$mode objfpc}
type
   TMyEnum = (two=2, five=5);
var
   E: TMyEnum;
begin
   E := TMyEnum(3);
   Writeln(E); // RunError (107)
end.

(From the above example one would say they are not valid.)


>> Furthermore I strongly disagree with your (2) statement. On the 
>> contrary, if there are all explicitly defined cases covered and there 
>> is an else statement (like in the TestRanges program above), a 
>> warning about the unnecessary case statement is very needed because 
>> it is an "implementation detail" of the compiler if the else block 
>> will ever be executed or not (AFAIR you supported this).
>
> The TestRanges program is not about an implementation detail, but 
> about undefined behaviour. Those are two completely different things.

There was a misunderstanding. I thought you meant with "implicit" values 
the "implicit default values" (=0):

program Project1;
{$mode objfpc}
type
   TMyEnum = (two=2, five=5);
   TMyObject = class
     E: TMyEnum;
   end;
var
   O: TMyObject;
begin
   O := TMyObject.Create;
   Writeln(Ord(O.E)); // R has an implicit default valuethat is invalid
   ReadLn;
end.

But you meant the implicit valid values (the holes). Now it's clear.

Best
Ondrej