Re: This doesn't look right

[Rik]

On 01/25/2016 12:16 PM, John W. Eaton wrote:
> On 01/25/2016 02:59 PM, John W. Eaton wrote:
>> On 01/25/2016 01:00 PM, Carnë Draug wrote:
>>> On 20 January 2016 at 00:40, Rik <address@hidden> wrote:
>
>>>> The second oddity is in event-queue.h
>>>>
>>>> class
>>>> event_queue_safe : public event_queue
>>>> {
>>>> private:
>>>>
>>>>    static void gripe_exception (void);
>>>>
>>>> This private function has a prototype, but no implementation
>>>> anywhere.  If
>>>> this is really meant to be a do-nothing function then it should at least
>>>> have a comment to that effect.  Anyone know why this is the way it is?
>>>>
>>>
>>> hg blame tells me it was added in 1054ab58cd58 [2] but again, there isn't
>>> many comments.  However, looking at the rest of the commit it seems to
>>> be copied code from unwind_protect_safe (compare with [3]) so maybe it
>>> was
>>> just a copy and paste mistake?
>>
>> Yes, it looks like a copy and paste mistake.  I'm pretty sure we don't
>> want exceptions to be thrown when we are already inside a destructor, so
>> I guess this should just be a warning, and we should change the one in
>> unwind_protect_safe as well.
>
> I don't understand why there isn't a link error if there is no
> implementation of gripe_exception.
>
> I'm not sure why these functions are declared static.  That doesn't seem
> necessary in this case.  Or that they even need to be members of the
> class.  It seems to me that we could just have a generic
> "warn_unhandled_exception_in_destructor" function.
>
> It does seem better to me to at least issue a warning rather than
> silently ignoring the exception.  But maybe there is a reason to silently
> ignore it instead?  I don't know.
>

Maybe put in the warning and see if it ever gets triggered?  We could make
make the warning fairly specific that if you see this you should report it
on the bug tracker or the mailing list.

--Rik