Return to Bizarro world

[tmcole]

Here's a strange error:


Debug Error!

Program: c:w2.......

DAMAGE: before Normal block (#0) at 0x11A9678

(Press retry to debug the application)


Anyone have a clue as to what is going on? Thanks.

[tmcole]

I've tracked this down further. The problem comes when I add the following code, which is perfectly okay (I apologize for the lack of indentation - can't figure out how to post code):

IF (ZMIN(JW) < 0.85*H(KT-1,JW) .AND. KT /=2)
KTWB(JW) = KT-1
DO I=IU-1,ID+1
Z(I) = H(KT,JW)+Z(I)
HKT2(I) = H(KT,JW)-Z(I)
H2(KT,I) = H(KT,JW)-Z(I)
END DO
END IF

This causes the following code to generate an assertion failure:

DEALLOCATE HKT2

Any thoughts? I tried finding "assertion" and "assert" in the documentation, but couldn't find anything.

[Steven_L_Intel1]

I'm guessing that you're using Intel Fortran, since I've never seen a CVF error look like that.

Any time you get an "assertion" error from a compiler, that means it's a compiler bug and you should report it, with an example, to the appropriate support channel.

Steve

[Jugoslav_Dujic]

To me it sounds like a mixed-language project where the host is in VC++ -- am I right? I've seen similar (although not identical) messages in such a setup. Do you have array-bounds checking on (but note that it doesn't work on assumed-shape (*) arrays)?

Jugoslav

[tmcole]

FYI, I've tracked it down to an array bounds error. I would always leave array bounds checking on, except that my program takes 5 minutes to compile with bounds checking on and 1 minute with it turned off.

Steve, at one time you were going to look into the cause of the different compilation times, but all I ever got back was that basically it was a complicated issue. Any effort being made to find out why code takes so much longer to compile with bounds checking turned on?

[Steven_L_Intel1]

At this point, we're not looking at compile-time issues for CVF - Intel Visual Fortran has a lot that is different "under the hood" and I would not expect the same problem to surface there. If you find you still have a compile time problem with Intel Visual Fortran, let us know.

Steve