Hello All,

I am attacking a program (F77) that does selection of records from a detail
set based on grouped keys that span multiple fields in multiple datasets.
All sets with indexes in the group are LINKed to an auto-master. Please see
the partial installation job below. The set I am retrieving records from is
the ORDHEAD set, but the values used in the search can match any index in
the G1 group.

TABLE CUSTORD TRANSPARENT
;
TABLE ORDHEAD LINK TO CUSTORD
CREATE SALES-ORDC MULTIPLE FROM SALES-ORD OPTIONS RC G1
CREATE BUYER MULTIPLE OPTIONS RC G1
CREATE CUSTPO MULTIPLE OPTIONS RC G1
CREATE CUSTDEPT MULTIPLE OPTIONS RC G1
CREATE CUSTOMER MULTIPLE OPTIONS RC
CREATE NO-BOS MULTIPLE OPTIONS NP RC
CREATE DATE-INVOICE MULTIPLE OPTIONS NP RC
;
TABLE ORDLINE LINK TO CUSTORD
CREATE ITEM MULTIPLE OPTIONS RC
;
TABLE ORDCOMNT LINK TO CUSTORD
CREATE DESCR MULTIPLE OPTIONS RC G1
;
TABLE ORDEXTN LINK TO CUSTORD
CREATE SALES-ORDC MULTIPLE FROM SALES-ORD OPTIONS RC
CREATE JOB-NUMBER MULTIPLE OPTIONS RC G1
CREATE SALES-LINE-IC SORTED FROM SALES-ORD LINE
;

The code I am modifying has worked great so far, but now, the users want to
see the ORDHEAD records in (I assume descending) order of the DATE-INVOICE
field.  Sorted chains will not help, because we are retrieving records with
many different Image key values.

Sure, I can sort an array or temp file to get the record numbers in the
desired order, but I'm on a quest to do this with Omnidex.

I know there is a technique in Omnidex to say "from the subset of records
already selected, return the records that match this other criteria on a
different key".  So, maybe I can add an IMSAM key on DATE-INVOICE and do it
that way.(???) Gee, this was sooooo easy with Speedware! (4GL)

But I'm getting a headache trying to deduce this from the Omnidex manual.
Amy help?  Any online reference that is easier to use than the Omnidex
manual?

TIA

-dtd