forums.silverfrost.com

[Dick Hearsey]

(1) I installed FTN95, Express version. I created a project consisting of 3 .for fixed-format Fortran files and 4 .inc include files. I got the 3 and 4 files shown under Source and Include in the project. When I attempt to build, 2 of the .for files compile, the 3rd which references the 4 include files fails: "Could not open C:\...(path)...\comoncpd.inc". The path quoted was correct (and the same as for the .for files). What went wrong? I was trying for a win32 .exe. OS: Windows XP SP3

(2) I removed FTN95-Express and installed FTN95-Personal. I repeated the project creation above in Plato, with the same result. Again, what went wrong.

(3) I was able to compile the program using the Command Line. I made 2 test runs with similar but differing results. The input data file was read and listed on the "print" file, unit # 2. The main calculation ran as I expected. However, soon after execution commenced, output intended for the screen, programmed "write(*,25)'NLOOP=',i3,' ... ", was written to the print file. Both runs then failed before all ouput was generated. Case 1 registered a "unit not opened or preconnected" failure. Case 2 registered "Read beyond EOF"; 'no more Reads' should have occurred. Both cases run satisfactorily with other compilers. Yet again, what went wrong?

Thanks.

[PaulLaidler]

With Plato you only need to add the source (.for) files to the project. The include files are automatically added by Plato. It may be the same for Visual Studio.

[Dick Hearsey]

I tried PLATO3 again, and got it to open the INCLUDE files, and hence compile and link. However, it does access them where the "Include files" box lists them, but in another folder, up one level. It will not work with copies only where PLUTO states it will look. Not good.

I have solved(?) the problem of placing screen (standard unit) on unit 2 (my listing file), by replacing all WRITE(*, occurrences with WRITE(6, . I also had to replace all READ(*, occurrences with READ(5, . In both cases the program would start behaing properly, and then use the *-unit incorrectly. Even worse!

Has anybody else had this problem?

Thanks.

[PaulLaidler]

In your program you can use relative paths for the include files. A file name without the path part would mean the file is in the same folder as the source file. You can use sub folders in the normal way. If you want to go up a level then ..\ may work. Failing that you will need to use the full path.

[Dick Hearsey]

When you say "in your program" are you referring to the Fortran source or the SilverFrost project?

I have always stored the include and .for files in the same folder. My references to the include files in the source code simply use the file-name with no preceding path.

To make PLUTO work I had to put copies of the include files in the folder that PLUTO quoted in my first project-build attempt where I had only defined the .for file names and locations. That was when it stated "Can't open include file path\name"; I used that path to locate where I put the copies. Now it shows "my" location, but reads from the PLUTO location.

[PaulLaidler]

From what you say, Plato should work as you expect.

Start again with a new project and make sure that the project file (*.ftn95p) is created in same folder as your primary source files.