topic /check:format appears to in Intel® Fortran Compiler

Peculiar PRINT statement behavior

William_S_1 — Thu, 06 Feb 2014 21:34:47 GMT

The subroutine I am uploading has an anomaly I can't explain.

Around line 61 there is a PRINT statement for the array QU(0:NQU)

It does not print the array, just the opening character string.

I checked with the debugger, and the value NQU is just as it should be, a positive number.

furthermore the other arrays DO print out as desired -

using the same Format statement.

So I am wondering - what is special about that array that would trigger this problem?

BTW, I have a special reason for dimensioning the arrays the way I do,

starting with 0 as the first element, but I am wondering if that "fools" the compiler.

I tried taking the array QU out of the calling sequence, but that had NO effect.

I can put everything into a

William_S_1 — Thu, 06 Feb 2014 21:52:43 GMT

I can put everything into a special project if you need to RUN it.

How do I upload a whole project at once?

ZIP the project folder and

Steven_L_Intel1 — Thu, 06 Feb 2014 21:57:37 GMT

ZIP the project folder and attach that.

OK, lets hope I did this

William_S_1 — Fri, 07 Feb 2014 02:35:10 GMT

OK, lets hope I did this right - -

The PRINT statement is at line 53 of the routine IDIV.

These routines are supposed to simulate long division with integers,

so its a package that does operations with very large integers.

You can put a read(*,*) to make it stop at line 54.

Your ZIP doesn't include the

Steven_L_Intel1 — Fri, 07 Feb 2014 17:02:10 GMT

Your ZIP doesn't include the sources, and you didn't attach all of them.

I zipped the entire folder

WSinc — Fri, 07 Feb 2014 22:24:57 GMT

I zipped the entire folder with all the sources in it, and when I moved into another computer, all the sources were there.

so I don't see why you say that.

The only Fortran source files

mecej4 — Fri, 07 Feb 2014 22:30:57 GMT

The only Fortran source files in your zip archive are the *genmod.f90 files generated by the compiler. There is not a single original source file in the zip from which those interface checking files could be generated.

List the names of your original Fortran source files, and we can check.

OK, here it is again -

William_S_1 — Fri, 07 Feb 2014 22:50:00 GMT

OK, here it is again -

I think some of the files were attached to a project in another folder.\

This one should have 7 source files in it.

You can see where I printed the number NQU, but the array QU does not print.

Thats around line 53 of the routine IDIV.

QU(0:NQU) should print out NQU+1

quantities, right?

I also tried ( QU(KK),kk=0,NQU )

but got the same result.

You've got two character edit

IanH — Fri, 07 Feb 2014 23:38:07 GMT

You've got two character edit descriptors (i.e. two letter A's ) in the format specification (labelled 102), but you don't have two character things in the item list in the print statement.

OK, I see where I forgot to

William_S_1 — Sat, 08 Feb 2014 23:05:08 GMT

OK, I see where I forgot to put the BLANKS in -

But why didn't it flag an error when I tried to print an integer

with an A descriptor?

Because we allow that to work

Steven_L_Intel1 — Sun, 09 Feb 2014 01:43:49 GMT

Because we allow that to work by default - some people used such a feature. There is an optional check for format mismatch you can turn on.

The mismatch isn't detected

IanH — Sun, 09 Feb 2014 06:39:12 GMT

The mismatch isn't detected with /warn:all /check:all /stand. is something else required?

/check:format appears to

mecej4 — Sun, 09 Feb 2014 12:08:00 GMT

The option /check:format appears to catch some mismatches between format descriptors and I/O list elements.

[fortran]program tfmt
integer :: i = 90 ! 'Z'
character(len=1) :: cz = 'Z'
!
write(*,'(1x,A5)')i !no error
write(*,'(1x,I8)')cz !flagged
!
end program tfmt[/fortran]