Re: nullablity

[Phil Clayton <phil clayton lineone net>]

05/10/15 21:49, Christoph Reiter wrote:
> On Mon, Oct 5, 2015 at 10:20 PM, Jasper St. Pierre
> <jstpierre mecheye net> wrote:
> > So, there's a few complications here. The first is that our public
> > APIs tend to have g_return_val_if_fail (..., NULL); even if things
> > aren't normally null. In that case, what does a binding do? Crash?
>
> If the NULL is only returned in case of an error (wrong usage, no GError, ...)
> the bindings could raise an exception pointing to the root cause instead
> of silently returning a NULL equivalent.

In writing bindings (for SML) I have been agonizing over whether a null 
value should cause an exception when introspection annotations say that 
the value cannot be null (in the absence of GError raised).  Is it more 
useful in practice to let, say, a signal handler blunder on rather than 
prematurely return due to a Null exception?  Ideally any function that 
can fail would have a GError.