RE: CEEDOD problem -- more info

["Joe Pluta" <joepluta@xxxxxxxxxxxxxxxxx>]

> From: Bob Cozzi
>
> Is your thinking that a data structure is fundementally different (on a
> procedure call) from a character variable?
> Does anybody have a V5R1 machine on which they can test a parm that is
> defined with LIKEDS() and see if the opdesc thing works?

I believe it was Hans who informed us that the big issue with data
structures is that there was originally some thought to providing the opdesc
information for all of the subfields.  This would obviously be a far cry
from treating the DS as a single character field, and would require some
interesting manipulation of the CEEDOD API.

At the same time, it might be nice to haev at least a compiler switch that
would force data structures to be described as character fields.  It might
make the programming a little more consistent.

Personally, I don't use opdesc.  I prefer to have all my parameter
information explicit.  That's just from long years of experience.  Opdesc
comes very close to assumption, and you know what happens when you assume...

Joe