Hello Harald,
the attached patch fixes inconsistent handling of passing derived type actual arguments to scalar dummies with VALUE,OPTIONAL attribute. As suggested by Tobias, we should consistently pass a hidden boolean flag that indicates the presence or absence of the actual, similar to the case of intrinsic types. For more details see the attached.Regtested on x86_64-pc-linux-gnu. OK for mainline?
OK. Thanks for the patch!
