https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80666
--- Comment #6 from Jos de Kloe <kloedej at knmi dot nl> --- Thanks for your test results and views on this matter. > (1) PARAMETER has a very precise definition in Fortran and AFAICT this > definition (named constants) does not match your use in the above quotation. > Are you meaning dummy arguments? As I am using several programming languages, it happens that I mix up terminology, so yes in this case I was referring to "dummy arguments". >(2) If yes, there is no correlation between the order the dummy arguments and > the individual type declarations. The order of type declarations is only a > matter of coding style, provided they follow the rule "define before using". I guess this answers my question. The "order of type declarations" is the one that should be in the right order. So my first example code was invalid fortran95. I still am not sure where in the fortran standard this is defined, but then it is a rather difficult document to read. Having noted this in bugzilla is useful by itself I think. Now others can find it when searching for issues.