FATAL ERROR - possible out of order or missing USE

[tkibedi]

When I try to re-compile an old quick win application under 6.5A the following strange error is displayed:

I:TiborRsIccSourceRsicc.f90(44) : Error: This name does not have a type, and must have an explicit type. [ATOMSTR]
Report = FALSE.
^
(241) : Severe: FATAL ERROR - possible out of order or missing USE
Error executing df.exe.

The variable [AtomStr] is properly declared in a module (say MyMod), which compiles without any error. The module is referenced (Use MyMod) in the source code. It seems to me, the same error message is generated for each subroutine at the first statement (after the variable declaration). A number of similar projects, generated with CVF6.1, have been successfully re-compiled and re-linked on the same system, but I could not resolve this one. I tried to delete the project and re-create it from scarch, but it did not help. Is there a specific help on this type of FATAL ERROR? Any clue?

[sabalan]

Isn't a dot (.) missing before FALSE?

Sabalan.

[tkibedi]

Unfortunately not. The dot was removed by the spell checker and was overlooked at the final submission. The error message in the posted text has been wrapped, which may makes it difficult to read.

[Jugoslav_Dujic]

I'm just guessing, but I know about the cases when order of USEs and
PRIVATE attributes in them affected association of variables (even internal compiler errors in 5.0A IIRC). I guess you have something like

 
MODULE MyModule 
LOGICAL:: Report 
END MODULE MyModule 

 
MODULE AnotherModule 
USE MyModule 
PRIVATE 
... 
END MODULE AnotherModule 

 
SUBROUTINE Rsicc 
 
USE AnotherModule 
USE MyModule 
 
... 
END SUBROUTINE 

What might have happen is that PRIVATE keyword in AnotherModule has somehow hidden the declaration of Report from MyModule. I didn't quite catch up the exact rules, though I follow the rule of thumb that USE statements in cases when MODULEs USE each other should be listed "from root upwards in dependency tree" -- in the sample I gave it means

 
USE MyModule 
USE AnotherModule 

And AFAIK the compiler's behaviour has changed from version to version (at least from 5.0A to 5.0D). Btw, Steve, do you know where exactly in the docs are the rules for USEing modules? (Of course, I might be completely wrong about it).

Jugoslav

[Steven_L_Intel1]

This particular error is always a compiler bug. Please send us a ZIP file of the project to vf-support@compaq.com, along with a problem description. Use VF Reporter or fill out and attach BUGREPRT.TXT from the DF98 folder.

Steve