Re: Service program signature violations?

["M. Lazarus" <mlazarus@xxxxxxxx>]

Ken,

At 11/29/01 01:25 PM, you wrote:
> >  It's annoying at BEST, just plain wrong at worst (and probably the
> >majority of the cases!)  I also can't see how someone would see this as a
> >feature and what code it would break.
>
>Procedures are called by slot number, not by name.  If compiling a module or
>service program without binder language caused the order of the exports to
>change from the way it originally was, then you either would get signature
>violations or the wrong procedure being called.
>
>It wouldn't matter to us as I use binder language for everything, but it
>could break code for people who don't.

  One of the nice things about the /400 is that it has a lot of built in
integrity checks.  Level checks, system guaranteed access paths, etc.  As
mentioned earlier in the thread, using RTVBNDSRC can cause unwanted results
if the number of exports has not changed between the versions it will not
generate an error.

  It s/b an easy task for IBM to generate the binder source in the order
that the modules were added.

  -mark