Re: Opinion on the MTF mh_n1074 doc

Guillaume AUDEON (guillaume.audeon@cril-ing.fr)
Fri, 20 Feb 1998 19:19:45 +0100

Richard Houldsworth wrote:

> [...]
> As Java does not allow passing by reference, the issue above
> does not arise. Indirection though an Variable is resolved before
> the call. Output parameters are implemented using a setVariable API
> call on Variables passed as ObjectReferences. This method does have
> the advantage of eliminating the side-effects of changing a variable
> without notifying the MHEG engine.

This allow for an easier and clearer implementation, IMHO.
Indeed, the interpretation of ":IndirectRef" is the one that applies to
all other elementary actions, i.e. de-reference and pass-by-value.

> [...]
> The solution to issue 11 seems to make sense to me, but it
> shouldn't be interpreted as meaning that :indirectref MUST mean
> pass-by-reference. It could also mean dereference and pass-by-
> value, depending on the application domain.

But today the solution would allow to interpret ":IndirectRef" as
pass-by-reference, I'm afraid.
And this interpretation would apply to parameters of the Call/Fork
actions, and not to the others MHEG-5 actions.
A bit strange, isn't it?

Best regards,
Guillaume

-- 
Guillaume AUDEON - CRIL INGENIERIE
5 square du Chene Germain - 35517 CESSON SEVIGNE (France)
+33 2 99 38 43 43 - mailto:guillaume.audeon@cril-ing.fr