It’s again time to organise our next monthly call which will be in the week of June 26-June 30; please see the following doodle poll to mark your availability:
Together with Lenore Mullin I have finished a paper on the Mathematics of Arrays in combination with Fortran (see some old threads in discourse). Last friday I submitted it to Computing in Science and Engineering and now I have to wait for a decision
I think it would be nice to have a collection of these for reference.
One that I happened to look at is the description of f66 do loops. F66 do loops were not directly or specifically required to execute once, rather the range parameters were restricted so that a single pass was always executed for conforming code. The one-trip execution feature followed indirectly from these constraints. The loop in the example code
do i = 1,0
write(*,*) 'FORTRAN 66: ', i
enddo
write(*,*) 'Current value of i:', i
Would have been illegal for five reasons. 1) the do statement would have required an integer statement label; the unlabeled do with enddo was not introduced until f90. 2) the termination value m2 is zero. In f66, the m1, m2, and m3 values were all required to be greater than zero. 3) the m1 value must be less than or equal to the m2 value, a constraint on the programmer that is violated in this example. 4) the enddo statement did not exist in f66, it would have been a labeled continue statement. 5) upon execution of the loop, the loop control variable i is undefined, so the final write statement is referencing an undefined integer.
Many compilers would not diagnose and detect violations to the m1, m2, m3 constraints, so they would execute the loop with a single pass. But that was not required or specified by the standard, the standard simply stated the requirements which were violated by the programmers in these cases.
Of course, the write statements also violate f66 in other ways. 1) Character constants did not exist until f77. 2) List-directed i/o did not exist until f77. 3) The use of * to specify the default output logical unit was not defined until f77.
@RonShepard, thanks for this detailed explanation. I will adapt the text. By thew way, I just saw another use of ENTRY that might be a useful addition.
I am not sure if the document as it is now should grow into a catalogue or a bestiary of such old-fashioned idioms, but the examples subdirectory is patient enough …
Hi @Arjen , I am interested in reading your paper on the Mathematics of Arrays in combination with Fortran , can you please send me the paper on henilp105@gmail.com . Thanks
