I noticed that the Beta 8 oneAPI compiler (Version 2021.1 NextGen Beta Build 20200715) for Windows gets an ICE for an example in the Fortran 2018 standard "C.6.5 Simple example using events":
D:\Projects>ifx /Qcoarray t.f90 Intel(R) Visual Fortran Compiler for applications running on Intel(R) 64, Version 2021.1 NextGen Beta Build 20200715 Copyright (C) 1985-2020 Intel Corporation. All rights reserved. me_emit_sub_node: ERROR: Subtraction of pointers should be done in me_emit_aminus_node t.f90(14): catastrophic error: **Internal compiler error: internal abort** Please report this error along with the circumstances in which it occurred in a Software Problem Report. Note: File and line given may not be explicit cause of this error. compilation aborted for t.f90 (code 1)
Overall, I am VERY impressed with this compiler's F2018 support - I have tried a LOT of things and, other than the one thing called out as not yet implemented (REDUCE), it has all worked (save for this error).
I note that beta users are asked to report issues here.
@Steve_Lionel or someone from Intel can clarify because the driver shown in the original post is "ifx".
Intel's documentation here states, "The Intel® Fortran Compiler IFORT (BETA) in the Intel® oneAPI HPC Toolkit beta release includes a full implementation of the Fortran 2018 standard, except for the REDUCE intrinsic. However, these features are not available when using the -qnextgen compiler option to IFORT or when using the IFX compiler driver."
So should IFORT (BETA) have been used here instead of IFX?
Hmm - perhaps I misunderstood. I tested a lot of F2018 features using ifx and they worked. But indeed if I use ifort I don't get the ICE. So this may just be my error, though ICEs are always bad. Indeed, all three of the examples I posted compile using ifort. My bad!