Thread overview
[Issue 3233] final switch could skip bounds checking in release mode
Dec 19, 2020
ponce
Dec 17
ponce
June 10, 2015
https://issues.dlang.org/show_bug.cgi?id=3233

Andrei Alexandrescu <andrei@erdani.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Version|future                      |D2

--
December 19, 2020
https://issues.dlang.org/show_bug.cgi?id=3233

--- Comment #1 from ponce <aliloko@gmail.com> ---
For example:

See godbolt output in https://d.godbolt.org/z/7x3Ehd


--------------- switch.d ---------------------


enum LOL
{
    a, b, c
}

import core.stdc.stdio;
void dosmthg(LOL l)
{
    assert(l >= LOL.a);
    assert(l <= LOL.c);
    final switch(l) with (LOL)
    {
        case a: printf("hello");
        case b: printf("world");
        case c: printf("!!!!");
    }
}


----------------------------------------------


The final switch always has a default case that call core.internal.switch_error

I'm not really if this is a good thing or not.
On one hand, it's like a enabled-on-release assertion so a good idea.
On the other hand, it's one more test in a final switch that could assume all
values are inside range. Maybe it's better to err on the side of safe.

--
December 17
https://issues.dlang.org/show_bug.cgi?id=3233

--- Comment #2 from ponce <aliloko@gmail.com> ---
Please close this :)

--