Re: Additional Glib::RefPtr Safety Mechanism

[Rob Pearce <rob flitspace org uk>]

On 31/08/2019 16:03, Karsten Pedersen wrote:
> The biggest issue is that often many of them wouldn't experience any
> catastrophic problems at all. They wouldn't even know anything was
> wrong with their code. It is only when I would mark it and send their
> binary through Valgrind it would light up like a Christmas tree!
> Undefined behaviour is frustrating like that;)

True, but even the best hand-holding nanny-code won't catch all the 
undefined behaviour - some of it is design rather than implementation - 
so it would do them good to be left floundering with a program that just 
misbehaves without warning, to teach them that they will, at some point, 
hit a mysterious unfathomable problem that doesn't respond to simple 
static or even run-time checks. Then you can teach them about core dumps 
and why they're useful ;)