The attached file produces an I/O error with a positive I/O status of 126,
when using I/O routines calling other I/O routines containing statements with advance="no"
This worked, however, in ifort 16, and it also works with gfortran, NAG, and PGI.
Is there an update on this issue? For the moment this is really a showstopper as we cannot rewrite major parts of our code to circumvent this regression.
I finally found the corresponding internal tracking id, DPD200414919, and your associated Premier issue on this same defect. I confirmed there is a fix (using your test case) but the availability of that in a future 17.0 update is not clear so I've inquired about that. I will post again once I learn more.
(Internal tracking id: DPD200414919)