Le 06/04/2014 19:46, Tobias Burnus a écrit :
> Mikael Morin wrote:
>> Le 06/04/2014 18:05, Tobias Burnus a écrit :
>>> It is supposed to propagate the information whether any of the
>>> components ("c") has a derived type. However, I made a typo: It
>>> should be "&has_final2" instead of "&has_final".
>> gfc_is_finalizable couldn't be used?
>
> No that requires that gfc_resolve_finalizers has run to setup some
> variables - and we are in that function … If one tried to run it
> nevertheless, gfortran ICEs ;-)
>
Argh. great. What about the use_assoc thing? Why is that needed?