Making Microsoft Fortran Powerstation program work with ifort

[rean_duro]

I have a program made in Microsoft Fortran Powerstation 4. It compiles and works properly on that enviroment.

However I am very interested in making it work under Linux. When I compile the sources with ifort, I get plenty of "catastrofique errors" and does not work at all.

Any experiences with this?

Thank you

[Steven_L_Intel1]

If you are getting "catastrophic errors", that's a compiler bug. Please provide us with the source that causes this error and the command you are using to compile. If you don't want to attach it here, you can submit through Intel Premier Support.

[rean_duro]

I am compiling with

dda: $(PROGRAMS)

ifort -o dda -I./ $(PROGRAMS)

where programs is a list of .for files.

see two files attached. input.for is not included in PROGRAMS.

Greetings!

input.for

routine.for

[Steven_L_Intel1]

Thanks - I also need iddefn.for.

Which compiler version are you using? Can you show me the output from the ifort command?

[rean_duro]

apologies,

iddefn.for is same as input.for

i tried (unsuccessfully) to rename it for clarity.

i am using

ifort (IFORT) 12.0.0 20101006

Copyright (C) 1985-2010 Intel Corporation. All rights reserved.

Error output is quite long. Mostly of what i see is

--------^

IDDEFN.FOR(117): error #6236: A specification statement cannot appear in the executable section.

COMMON /STAT/ TEMP1(NNB,2),TEMP2(NN5),JSTP,IOCRL,ISTAT,

and then

catastrophic error: Too many errors, exiting

[mecej4]

You have many source lines without statement numbers that start before column 7. How did they lose the leading blanks? Did you convert source that had leading tabs?

Because of those lines, the compiler can get confused about which lines are declarations and which contain executable statements. Forget about whatever error messages you get now. Fix those lines first so that the input has the proper format for fixed source. Then, run the program through the compiler.

[Steven_L_Intel1]

I get no errors compiling the code. Remember that INPUT.FOR (and IDDEFN.FOR) is an INCLUDE file and should not be compiled separately. You will get that kind of error if you try to compile the INCLUDE file as a separate file.

[rean_duro]

All I had to do is to clear up some TABS and use the -align all and some other flags.

Now it compiles properly and builds a neat executable.

Thank you a lot.

[rean_duro]

Source has all sort of mixtures in shape and content. Happily it is working now. Thank you!

[Steven_L_Intel1]

Glad to hear it.

[Mazdak_P_]

Hi. I have a FORTRAN code which has been written in Visual Fortran 4.1, and right now I use Intel compiler under the Microsoft Visual Studio 2008. When I run the code with Intel compiler, it doesn't give me the correct results (cuz I have the correct results) while the compiler doesn't show any error. There are some questions:

1) is it possible that some of statements can not be compiled in Intel, since the code has been writeen in old version of FORTRAN?

2) I have heard there is feature in compilers that makes compilers to neglect errors and run the code based on some of it's default? if it's

true, how can i dissable that?

thanks

[Steven_L_Intel1]

Please post this question in http://software.intel.com/en-us/forums/intel-visual-fortran-compiler-for-windows and provide some more details. If you can attach a sample program that shows the problem, that would be best. Intel Visual Fortran supports pretty much all of the PowerStation syntax - you would get errors for things not supported. But it is often the case that coding errors can cause different results when changing compilers.

There is not really a feature as you describe it. Let us help you find out what's wrong.

[mecej4]

Does your old code require that local variables be SAVEd in order to work correctly? That is likely if you used Digital/Compaq Visual Fortran. If so, read about the /Qsave and /Qzero options.

You may also give some consideration to fixing up the code so that it does not rely upon such non-standard behavior.