Link Copied
At 17 it goes gray because it is inaccessible but not yet cleaned up by GC.
I cannot download test.zip it says, scanning for viruses. So please put the code up as a text block using the standard insert function. </> Function on second line of the tab bar.
I made minor changes to the code so I could single step through the code and not lose steps. This needed the entry put into a subroutine. And I added implicit none and declared the variables so I knew when they existed.
When I start the code and stop at first line
jnum is declared and is set to the standard open 0. and ctext is empty as expected.
Now we see text and integer filled in
then step into dummy
CTEXT1 is declared and empty, but jnum1 to 3 are not yet set at line 19
Teh next step is to line 24, which shows the declared varaibles set as zero except jnum that comes in as 1.
Line 25 sets a few variables
Step to entry dummy_2, and jnum2 is set to zero as declared.
now it steps to the subroutine, and jnum3 is undefined, but jnum2 is defined, it is a bit weird, the entry declares the varaible.
now the jnum3 is declared,
Normal variables, set to proper values,
more correctness
Even more correctness
now at entry jnum2 is undefined but known
jnum2 is declared and correct, jnum3 is undefined but known
defined.
Set all values.
but now we get garbage when it returns to the entry point.
Now the original
As expected
Now they are unknown to the subroutine
You are not at numbers yet so it knows it will need them, but has not set them.
jnum1 and 3 set to zero as expected and jnum2 is stuck at entry point as undefined, it knows it needs it but it is not set
JNUM2 is never reached
It never stops at 21.
PROGRAM CONSOLE
implicit none
CHARACTER CTEXT*(5)
integer jnum
JNUM = 1
CTEXT = 'Hallo'
CALL DUMMY (JNUM,CTEXT)
END PROGRAM CONSOLE
SUBROUTINE DUMMY (JNUM_PARAM,CTEXT_PARAM)
implicit none
integer jnum_var, jnum_param
CHARACTER CTEXT_PARAM*(*),CTEXT_VAR*(20)
CTEXT_VAR = CTEXT_PARAM
JNUM_VAR = JNUM_PARAM
PRINT *, CTEXT_PARAM
RETURN
ENTRY DUMMY_2
PRINT *, 'Hallo'
RETURN
ENDPlease use the
Button to make the code nice and easy to read, it is a thoughtful thing for the reviewers.
Your code does not have implicit none or the numbers declared and as such you observe what you observe, but if I bring the code up to modern standards with implicit none then I get
Correct behaviour. It has been many years since I failed to use implicit none, it is very dangerous coding not to use implicit none, when you fix old code like ULARC UCB 1973, you see the mistakes, and these are a problem.
The debugger works it does not have errors, but your code is obsolete, do not expect a modern compiler to fix coding issues well.
It should be an error in Fortran not to use implicit none - Steve???
Putting aside all of the interesting stuff we have discussed, there is a clear difference in IFX and IFORT as to how it handles passed arguments once the debugger is inside the subroutine and we want to look at the passed arguments, as was advanced in original posts, if you hit the circular arrows the data for the arguments disappears in IFX and not in IFORT.
I would at least expect the same functionality in IFX. Perhaps Intel will make the change or we just learn do not touch the circular arrows. The popup message is in my humble opinion, incorrect, the values are known as they are passed.
This is the crux of the original posts.
