Strange Linker Errors in IVF 11.1.048

[jkrob]

Hello,

I have made the leap from CVF 6.6C to IVF 11.1.048/MVS 2008. I have converted over my project I have been building & updating over the last 10 odd years. Reading the CVF-IVF transition paper, I have gone through & converted all of the "USED DFLIB" entries to the "USE IFPORT" or whatever was required for the specific FORTRAN calls in the subroutines. Upon compiling/linking, I have come across some odd Linker errors which will not resolve themselves & I need help.The following subroutines/functions are causing LNK2019/2001 errors:

BEEPQQ

GETTIM

SYSTEM

SLEEP

Also, as for the GETTIM error, it was initially flagged as GETTIMI2with the arguments listed as INTEGER*2. When I changed them to INTEGER*4, the error changed to just GETTIM. The documentation says the arguments can be either I*2 or I*4...WHAT IS THE PROBLEM?!?!?

I have tried to configure the Properties to match the closest I had in CVF but I cannot understand why these few would not work & all others in my 543 subroutines do.

Thanks in advance,

Jeff

[DavidWhite]

Is "Generate Interface Blocks" turned on (Under Fortran / Diagnostics)?

You could try turning this of and rebuilding.

David

[jkrob]

David,

Hey - thanks for the reply. No, 'Generate Interface Blocks' is not enabled.

Jeff

[Steven_L_Intel1]

Please attach the buildlog.htm from a Rebuild of the project (that includes the compile and link).

[jkrob]

Steve,

Attached is the Build.htm you requested.

Thanks in advance,

Jeff

[Steven_L_Intel1 · Accepted solution]

I see you have /fpscomp:nolibs set. Any particular reason for that? Try disabling that (under compatibility > PowerStation). Please attach a ZIP of one of the .obj files from your compilation (any one will do.)

[jkrob]

Steve,

Heyyyy...that did it! It now compiles completely and error free. Now,the programcrashes when it executes...but it compiles error free (I know, small steps...).

Thanks,

Jeff

[jkrob]

BTW - since I have your attention ;-), is there a way to make the routines compile in alphanumeric order like it did in CVF? Watching the compile process, the source files seem to be chosen at random. It takes around 25 min to build the program on a Dell XPS system (not a slacker by any means) where, on CVF, it seemed to be *much* faster (I just tested...3 min!!!). If the source progress was alphanumeric, I could at least know (ballpark) how much longer to wait.

Also - the Traceback feature isn't working for squat. The program is crashing 3 subroutine branches off the main routine & all the usefull info Dialog gives is the error &the image. The Line & Source is *unknown*...that's real good.

Jeff

[Steven_L_Intel1]

CVF did not always use alphabetic order. The order is partially determined by the dependency graph created by USE and MODULEs. You don't have any direct control over the order. The dependency generator that Intel Fortran uses is more reliable than the one CVF used, which tended to get confused by deeply nested dependencies.

I don't know why you aren't getting a traceback. The log you provided earlier indicates you have the proper options for that set. If you'd like us to investigate, please attach a ZIP of the project (after a Build > Clean) and instructions for how to reproduce the error.