Re: [gst-devel] g_return_val_if_fail usage in dparams

[Steve B. <st...@st...>]

I have no problem with this change.

cheers

On Mon, 2005-01-03 at 20:32 +0100, Stefan Kost wrote:
> hi hi,
> 
> even though the current dparams will be replaced, the current implementation has
> something I consider as a flaw.
> No only incomming parameters are checked with g_return_val_if_fail(), also
> during the method this used. The problem is that during unit test this causes
> the test to fail, without that one can check for it.
> 
> The specific case is dparammanager.c::gst_dpman_get_dparam().
> If name is nonsense a method that is invoked will return NULL and therefore the
> method will abort with a CRITICAL instead of just returning NULL, so that the
> callee could handle the problem.
> Infact GError could be used to give the callee detail what has caused the error.
> 
> To avoid API changes, I would like to change those
>   g_return_val_if_fail(dpwrap != NULL, NULL);
> statements into
>   if(!dpwrap) {
>     GST_INFO("invalid dparam name");
>     return(NULL);
>   }
> 
> Any objections? Steve?
> 
> Ciao
>   Stefan