SQL Field Defintions

I brought this up on a previous post about doing data structures to be
used with a SQL Select statement. 

If I have something like this:

d dsMODATA...
d                 ds                  Qualified
d   CITEM...
d                               15a
d   CITWH...
d                                3a
d   ACSTD...
d                               15p2

Has anybody come up with a good way to deal with this? It seems like I
have four choices in dealing with this.

1. Lookup the field type and size in the MODATA file and hardcode. Ugly.

2. Define the file in the F specs but never open. Reference the field
from the dummy file. Real Ugly.
3. Define a qualified external data structure for MODATA and base it on
a pointer so it takes no storage. Reference the fields from the
qualified data structure. Not as ugly but fills up compile listing with
a ton of junk.
4. Define am empty table for each data structure and reference as
external data structure. Really ugly.  

Anybody come up with a better way of doing this and no I don't want to
use select * from? 

The obvious solution would be:

d dsMODATA...
d                 ds                  Qualified
d   CITEM...
d                                     ExtFld(CITEM:MODATA)
d   CITWH...
d                                     ExtFld(CITWH:MODATA)
d   ACSTD...
d                                     Extfld(ACSTD:MODATA)

or

d dsMODATA...
d                 ds                  Qualified
d   CITEM...
d                                     RefFld(CITEM:MODATA)
d   CITWH...
d                                     RefFld(CITWH:MODATA)
d   ACSTD...
d                                     RefFld(ACSTD:MODATA)

But IBM doesn't do that.                    

I suppose one solution would be to use my own pre-complier but that is
ugly too.

I guess the best solution is a dummy data structure but just wondering
if someone thought of a better way.