I agree. The compiler that is confused is just an hypothesis, and it would be a compiler bug.
Thanks for these answers.
Indeed, removing USE grid_ames does not solve this issue.
Would you have some information about bug related to the migration from pgf90 13.10-0 to pgf90 (aka nvfortran) 21.9-0 ?
Indeed, as explained previously, the compilation works well with pgf90 13.10-0 but not with pgf90 (aka nvfortran) 21.9-0, so it seems to be related to the compiler version.
Note once again that if you were posting a true MWE, we could confirm if it’s a compiler bug or not. At the point we can just say that it looks like a compiler bug.
Would it be possible to define which details would be expected for this MWE ?
In the case of this topic, a code for which the only error reported by the compiler is the one that you don’t understand. When I tried compiling the code you have posted, I got a lot of errors that were not related to that.
Are you using pgf90 13.10-0 and pgf90 (aka nvfortran) 21.9-0 as specified for this compilation ?
No, I am using gfortran or ifort, but that’s not the point. The code you have posted contains obvious errors, such as type definitions after contains, etc (I’ve listed them in a previous message) and I doubt you could compile it at all even with pgf90 13.