Re: C is 'better' than Ada because...

[kennedy1@bwmail1.hcc.com]

In <31f7db04.80562082@netline-fddi.jpl.nasa.gov>, kdq@emoryi.jpl.nasa.gov (Kevin D. Quitt) writes:
>No, that's like C's switch statement.  COBOL uses self-modifying code for
>computed gotos.

No it doesn't.  COBOL's <GO TO paragraph-1, paragraph-2... DEPENDING ON 
integer> statement works just like FORTRAN's computed GO TO statement.  (A 
given compiler might produce self-modifying object code to implement it; but 
that's neither here nor there; I have known hardware where self-modifying code
was the only way to return from a subroutine.)

You're thinking of the <ALTER paragraph-consisting-of-one-go-to-statement TO
PROCEED TO new-target> statement, which is the COBOL equivalent of a FORTRAN
ASSIGNed GO TO.  It is worse than the FORTRAN version, because there is no
clear evidence in the GO TO that it is ever altered, aside from the fact that
a paragraph consisting of only a GO TO is rare otherwise.  (Also, the
statement may read <GO TO.>, in which case it is clear to the reader that an
ALTER must exist.)

But even that isn't as horrid as the COBOL <PERFORM paragraph-1 THRU
paragraph-2> statement, which does a goto to paragraph-1 after inserting an
invisible, self-erasing goto at the end of paragraph-2.  Many's the time I've
hit code that took a "wild branch" when the program started a PERFORM THRU,
did a GO TO out of it (perhaps aborting a bad transaction), and later, while
processing a wholly separate transaction, happened to drop through to
paragraph-2, where the uncancelled invisible goto was still lurking.  It's
contemplating stuff like this that makes me recall how COBOL almost didn't get
an IF statement "because 'if' isn't a verb," or how no-one ever got around to
defining the results of the COBOL <COMPUTE variable = arithmetic-expression>
statement.