Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!csd4.csd.uwm.edu!bionet!ames!purdue!haven!adm!cmcl2!lanl!jlg
From: jlg@lanl.gov (Jim Giles)
Newsgroups: comp.lang.fortran
Subject: Re: Is this a correct Fortran 88 subroutine?
Summary: How about this syntax instead?
Message-ID: <14040@lanl.gov>
Date: 8 Sep 89 22:19:35 GMT
References: <790hallidayd@yvax.byu.edu>
Organization: Los Alamos National Laboratory
Lines: 37


>        SUBROUTINE X(L)
>        LOGICAL L
>        IF: IF ( L ) THEN
>            ELSEIF              ! This is not an ELSEIF statement
>            ENDIF IF
>        END

I have always favored a syntax in which block names _always_ appear
on the left:

        IF: IF( L ) THEN
           ...
        IF: ELSE
           ...
        IF: ENDIF

Here, the visual ambiguity is gone (there never was any computational
ambiguity).  Some people regard this is ugly, but if the purpose
of block names is as a documentation aid as well as a flow control
aid, then this syntax serves the purpose well.  All the block names
are easier to find on the left and will line up on properly indented
code.  Here is an example with loops:

        LOOP: DO I=1,1000
              ...
        LOOP: EXIT IF (COND)           !EXITS CAN BE CONDITIONAL
              ...
        LOOP: EXIT                     !OR UNCONDITIONAL
              ...
        LOOP: END DO

Of course, CYCLE and CYCLE IF statements are also possible.  Note the
with the block name on the left, there is never any doubt about which
statements on the page have a flow control effect - and no doubt about
which construct they belong to.  With the block names on the right,
they are effectively buried in the middle of the program text.