Re: Global variable search

[Rory Hewitt <roryhewitt@xxxxxxxxx>]

Kurt,

Too true! I actually started writing a parser for exactly this, but I got so
bogged down, I gave up.

One of the 'worst' things I've seen was where both the main PR and PI were
in the program source member (i.e. the PR wasn't in a copybook), but they
were separated, with the PR at the very top of the D-specs and the PI at the
bottom, just before the C-specs. Of course some people probably like this
technique and think mine is awful. C'est la vie.

Having said this, I've written programs where both the main PR *and* the
main PI were in a copybook - nothing in the main source. It was my attempt
to put all interface-related stuff into a single copybook, which can also be
used by other programs to call my program. You can do it, but it ain't
pretty - lots of conditional compilation stuff. But you'd never know (unless
you parsed all the copybooks) what the program interface was!

Rory

On Tue, Aug 4, 2009 at 9:44 AM, Kurt Anderson
<kurt.anderson@xxxxxxxxxxxxxx>wrote:

> Rory said:
> "I'm not saying that my method is 'correct' (or even 'better' than anyone
> elses!), but there is a reason why companies that write source analysis
> tools can charge the big bucks - it's a lot more complicated than it might
> at first appear..."
>
> Definitely.  And after I read that, I suddenly realized there's a huge
> glaring error in the last numbered idea I had - and that was the fact that I
> didn't take into account other prototypes that were unrelated to the main
> procedure.  D'oh!
>
> -Kurt