Migration Compaq to Intel Fortran

[Guilherme_S_]

Hi, 

I´m migration an application Compaq to Intel Fortran...

After extraction the code in Visual Studio, the only build error I found was a "dot" in one funcion call. (I think the print screen attached can provide a better understanding ) Removing the "dot" in Intel Fortran was enough to perform a successful build.

What I really don't know is what this syntax means in Compaq fortran and why the compiler did not sign that line as an error. I'm afraid that removing this line may remove some of the applications functionality

[Steven_L_Intel1]

The dot has no meaning in this context in Compaq Fortran and I would have expected CVF to give an error if that dot was in the statement field. Can you attach a sample source that compiles ok in CVF but not in Intel Fortran? What was the file name (did it end in .f90 or .for or .f?)

[Lorri_M_Intel]

It's probably not an actual "." ... it is more likely an unprintable character that CVF ignored.

The underlying code to read a users' file changed between CVF and Intel Fortran, and it's possible that something that previously ignored is no longer ignored.

     --Lorri

[Steven_L_Intel1]

Guilherme sent me the actual source, and to my astonishment, the character really is a normal ASCII period. Nevertheless, that's not legal Fortran syntax (where it was used) and if CVF didn't detect an error, that was a CVF bug. Dots have been a sore point for this compiler's parser forever, and I suspect this is just one aspect of that. We have fixed a number of bugs in this area over the years, so this was another bug we fixed, allowing the error to be reported.

Intel Fortran should compile any legal source that CVF could compile. It will, however, detect MANY errors CVF didn't.

[Guilherme_S_]

thank you guys!

ps: google usualy pays $1K dollars for every bug an user finds... :-)  just kidding...

[Steven_L_Intel1]

So you are going to pay us $1K for each of those dots?

[Guilherme_S_]

hehehe, I´m not that rich :-)

[lklawrie]

Wonder if the dot gets confused in the parser because it's one of the "alternatives" for derived type %?  or the old "Structure" extension.

[Steven_L_Intel1]

Yes, that's the aspect that has given us many headaches over the years. Haven't seen a new one in quite a while, though.