Fortran Compiler 2025.3.1 crashes

[Kai8]

during a "build all", the compiler freezes in a specific subroutine and does not come back again. Working in Visual Studio, I can only abort the process.

Exiting, re-starting VS and doing a "clean all" does not help, the compiler will stop in the same subroutine again without any message.

[Kai8]

I forgot to mention that I encountered this problem only after the update to 2025.3.1. There were no issues with the previous versions.

[andrew_4619]

Do you only have one subroutine per file then? Or did you mean it always stalls on the same file?

Does the file have lots of complex dependencies or can you make a simple reproducer that others can test?

[Kai8]

There are 3 subroutines in this file and I cannot determine on which one it fails. They are connected to modules, which makes it difficult to extract and share.
One of the modules, which define the variables, was changed before the this happened. I assume there might be a not-defined variable. But I do not get an error message and cannot say for sure.

[andrew_4619]

I presume you tried deleting each of the three subroutines in turn to see if the file compiles OK?

[jimdempseyatthecove]

FWIW I ran into a similar problem, it may be related to  yours (or not)

My issue related to a module with contained procedures

.AND.

where a contained procedure made a reference to another containd procedure in the same module

.AND.

the other contained procedure was a forward reference (i.e. followed the referencing contained procedure)

.AND.

the other contained procedure has an error

My solution was to move the erroring procedure in front of the referencing procedure.

This results in a non-crash (hang) and with error message display (to the original error causing the issue).

Jim Dempsey

[Kai8]

thanks for your comments.

I resolved my issue meanwhile. It was less complex than the one described above: Initially, a vector was defined and allocated in a module and was used by the subroutines, but it was removed from the module later. One of the subroutines was still using it:

USE MODULE **bleep**, ONLY: BadVector   -> where BadVector did not exist in this module anymore

That seems to have caused the compiler to hang.

To my view these are essential things a compiler should handle with a proper error message, i.e. that the variable has not been defined.
I am always using IMPLICIT NONE in my codes. I could have repaired it immediately, but this cost me a day of work.

I hope intel can repair soon.

[andrew_4619]

I make that error due to bad spelling all the time and get:

error #6580: Name in only-list does not exist or is not accessible

I think there must be some other feature(s)  of the error that causes the hang. If it has some complex subtleties  you are unlikely to see a fix and time soon without posting a reproducer for Intel devt to work on.

[Igor_V_Intel]

In the trivial example replicating your scenario, I can't see the hang. I am getting the expected error from the compiler

error #6580: Name in only-list does not exist or is not accessible. [BADVECTOR]

So, would be great if you can share more details with some code snippets. I believe there is an issue with integration to VS on your machine.

[Kai8]

yes, there is some other feature!
Meanwhile I observed this hanging of the compiler in several situations and with different programming errors not related to a module. I guess now, that it has something to do with the integration into Visual studio. The errors are reported in the buildLog.htm after I had to stop the compiler manually. 

My system worked before I updated to  2025.3.1. two days ago. Either the updating script did did not integrate it correctly, or the new version itself has a flaw.