error #5082: Syntax error, found INTEGER_CONSTANT

[Peter_Eckhoff]

I am trying to recompile an old Fortran 77 fixed format program.  There are revision date codes in columns 73 to 80 that are generating the error message below.  I have gone into Project | program Properties | Fortran | Language  and tried **ALL**  combinations of Source File Format and Fixed Form Line Length settings to no avail.  I have changed the extensions from *.for to *.f and still no solution.  I have looked online but did not find a solution.  Here is a sample code snippet and the resultant error messages:

CC

      dimension vmx50(50,4),imxdt(50,4),vmxxx(50,4),vmxyy(50,4),        110198

     *          nmx50(4),mspac(2000),vmxhi(20000),jmxdt(20000),         110198

     *          vxbuf(128),ixbuf(128),ifnxh(4)

CC

1>Compiling with Intel(R) Visual Fortran Compiler XE 13.0.1.119 [IA-32]...

1>C:\OBODM\OBNU\ROUTA.f(22): error #5082: Syntax error, found INTEGER_CONSTANT '110198' when expecting one of: <IDENTIFIER>

Using:

Microsoft Visual Studio 2010

Version 10.0.30319.1 RTMRel

Microsoft .NET Framework

Version 4.0.30319 RTMRel

PS I tried to change the fonts above to Courier New and others but the transfer from Word to here would not allow anything other than Calibri.  Suggestions would be appreciated. 

[TimP]

No doubt you've looked at the options summary relating to this question. The default invoked for either .f or .for should be /fixed /extend-source:72, and any text beyond column 72 would be ignored. If you have somehow ended up with /extend-source:80 (or 132) or /free (or the older versions of those), you will have trouble with sequence numbers or revision dates in that part of the data field. I recognize the possibility that the options don't work as documented, but that would be a bug. I think I still have awk scripts to remove those numbers (yes, they work with Windows gawk).

[Peter_Eckhoff]

Thanks for the quick feedback. I just looked at the Compiler section and this is how it read: /nologo /debug:full /Od /fixed /warn:interfaces /module:"Debug\\" /object:"Debug\\" /Fd"Debug\vc100.pdb" /traceback /check:bounds /check:stack /libs:static /threads /dbglibs /c This is with the Fixed Format and 72 character line set in the Language section of the Properties. I could GAWK it but then I would lose the change dates that are keyed in the source code and other documentation. I think that this is a bug. Do you know if Intel reads these problems and escalates them to bug status or do we have to proactively report this as a bug?

[mecej4]

I took the three lines of code above, added a "Program xxx" before and "End" after the lines, and ran the compiler with your options. The Windows compiler13.0.1.119 did not display any error messages. If there is a bug as you say, you have yet to post source code that really triggers the bug.

[Steve_M_3]

I am having the same problem. I started with an existing file, with stuff in column 73 on many lines. Got same error when trying to compile. I went to one line that was triggering the error, deleted the character in column 73, error for that line went away when I compiled. I typed the character back into column 73 on that line, and the character showed up in green (visual studio now sees it as a comment). And sure enough, when I compiled again, the compiler had no problems with that line. I got 40 files, 50000 lines of code, and 6 different versions of this program. I don't want to have to manually do this...

[Peter_Eckhoff]

Hello MECEJ4, I followed your suggestion but for the whole subroutine. It did **not** retrigger that error message. I had looked for horizontal tabs and did not find any. I added a few more subroutines and the error reappeared in the original subroutine. I modified the source codes so that there are no warning or remark messages; only the #5082 messages. Only about 6 subroutines were processed before the compiler exited. I think I will compile each subroutine outside the IDE. Do you have any suggestions that might be better? If interested, the source code is available on line for download.

[mecej4]

Peter Eckhoff wrote:

Hello MECEJ4,

If interested, the source code is available on line for download.

URL, please?

[Peter_Eckhoff]

Hello Steve, I tried your suggestion and it did not work when the "sequence number" was on the same line as a continuation line or a If Then line. There was no problem if it was part of a Comment line. With the IF then line, the number was green with a gray background but the compiler still flagged it. With the others, the number was still black with no gray background. Deleting and retyping the number made no difference. However, when I did what MECEJ4 suggested as a stand alone program, the problem with that set of sequence numbers went away. When it was brought back in as a subroutine, the problem reoccurred. This is old code from the late 80's early 90's. I think I will manually compile each subroutine to see if the problem is with the code or an effect from the IDE/compiler interaction. Hope you have a good weekend and thanks for your input. It was much appreciated.

[Steven_L_Intel1]

Try turning off /warn:interfaces. We've seen a bug parsing fixed-form source in some cases where that is enabled.