Did the rules for do-construct-names change?

[rwg]

With compiler Version 11 statements like the following compiled without any warning:

DO_W1: DO WHILE(INDN.GT.0)
....
ENDDO DO_W1

After upgrading to Version 12.0.4.196 I get following error message:

error #5082: Syntax error, found '(' when expecting one of: =

Did the rules for do-construct-names change?

It seems that all do-construct-names beginning with DO will give this error message.

[jimdempseyatthecove]

Haven't tried this but try removing the underscore

DoW1: ...
...
ENDDO DoW1

Jim Dempsey

[rwg]

Even

DOI:...
ENDDO DOI

fails. I can compile the file without any warning after changin DO_W1 to DD_W1 .

Seems to be a bug in Version 12.04.196.

Ralf

[Steven_L_Intel1]

Are you using /names:as_is? I know of an error message that would result if you used "do while" in lowercase and /names:as_is. The paraphrase you show does not run into that problem.

My guess is that your actual code is different.

[rwg]

Hello Steve,

here are my compiler options:

/nologo /debug:full /Od /DDEBUG /debug-parameters:all /warn:declarations /warn:unused /warn:truncated_source /warn:uncalled /warn:interfaces /Qauto_scalar /align:commons /assume:byterecl /Qfp-stack-check /iface:cvf /module:"c:\sw\lib_entw\modules" /object:".\Debug/" /Fd".\Debug\vc90.pdb" /traceback /check:pointer /check:bounds /check:uninit /check:format /check:output_conversion /check:arg_temp_created /libs:static /threads /dbglibs /c

And her my input and output:




Remove the O from DO_W1 and it will compile without any error

Ralf

[Steven_L_Intel1]

Ok. This is a .f and fixed-form source. I missed that part. I can reproduce the problem that way. I found that this issue has already been fixed for a future release, but I don't expect to see the fix for a while. The issue ID is DPD200156415. The workaround is to not have a label that starts with DO.

[jimdempseyatthecove]

In Fixed Form column 1 with D in it are Debug lines (/d-lines) or comment lines (/nod-lines)
Tags (statement lables) should be used in free form (else expect problems like this).

Jim Dempsey

[Steven_L_Intel1]

D-lines aren't the issue here - the code has blanks in columns 1-6.

[rwg]

Thanks Jim and Steve.

This was one file of 8164 (1,5 million lines of fortran code without comments) which I recompiled after upgrading from Version 11 to 12. The files contain pure F66 up to F2003 extension and I could not find any further error. You did a good job. Thanks to you and your colleagues, Steve.

Ralf

[Steven_L_Intel1]

This was fixed in 13.0.