I understand the sentiments here, and indeed prior to the PC age, I took great pains only to write Fortran that would move easily from machine to machine, bearing in mind that one rarely had a choice of compiler – just the one that came with that particular machine, warts and all. In the PC age, under DOS, there was an embarrassment of choice, and like you, I ran my programs through every compiler that I could find for purposes that I might now describe as QA, but then I would have expressed in the language of portability.
Once one dips one’s toe into Clearwin+, however, one is back in the era of Hobson’s Choice. Suppose, for instance, that you have made an error in the Clearwin+ code (or there is an error in Clearwin+) you will never find it with a different compiler, because Clearwin+ is the same and comes from the same place whether you use it with Gfortran or FTN95. As you discovered, it’s easier with FTN95. It’s only the bits of your program that don’t rely on Clearwin+ that are worth testing with a different compiler.
My experience, such as it is, is that using Clearwin+ locks you into FTN95 as effectively as it locks you into Windows, and takes you back into that early era. However, there are consolations.
Firstly, the Clearwin+ parts of your program will be associated with data input and results presentation. In the former, you can easily tell if the input data has actually been input; and in the latter, especially if you present the results graphically, you can see they are presented correctly. Making an old Fortran program into a fully fledged Windows application will inevitably mean that the majority of the program falls into those two categories. You can test the bit in the middle with any compiler by simply extracting the relevant routines, feeding them a dataset from a file (which is what happens with File|Open anyway) and storing away the results in another file.
Moreover, data input can only function at the speed of the human user and therefore the processing speed that depends on the compiler tends to be rather immaterial. When it comes to just simply drawings and pictures at the end of the day, speed is again rather secondary and you can normally tell by inspection if the process completes correctly.
Eddie
