Joining to One Cross-Referenced Segment From Several Host Segments

In this section:

You may come upon situations where you need to join to one cross-referenced segment from several different segments in the host data source. You may also find a need to join to one cross-referenced segment from two different host data sources at the same time. You can handle these data structures using Master File defined joins.


Top of page

x
Joining From Several Segments in One Host Data Source

In an application, you may want to use the same cross-referenced segment in several places in the same data source. Suppose, for example, that you have a data source named COMPFILE that maintains data on companies you own:

The DIVSEG segment contains an instance for each division and includes fields for the name of the division and its manager. Similarly, the PRODSEG segment contains an instance for each product and the name of the product manager.

Retrieve personal information for both the product managers and the division managers from a single personnel data source, as shown below:

You cannot retrieve this information with a standard Master File defined join because there are two cross-reference keys in the host data source (PRODMGR and DIVMGR) and in your reports you will want to distinguish addresses and dates of birth retrieved for the PRODSEG segment from those retrieved for the DIVSEG segment.

A way is provided for you to implement a join to the same cross-referenced segment from several segments in the one host data source. You can match the cross-referenced and host fields from alias to field name and uniquely rename the fields.

The Master File of the PERSFILE could look like this:

FILENAME = PERSFILE, SUFFIX = FOC, $
SEGNAME = IDSEG, SEGTYPE = S1, $
   FIELD = NAME,     ALIAS = FNAME, FORMAT = A12,     INDEX=I, $
   FIELD = ADDRESS,  ALIAS = DAS,   FORMAT = A24,              $
   FIELD = DOB,      ALIAS = IDOB,  FORMAT = YMD,              $

You use the following Master File to join PERSFILE to COMPFILE. Note that there is no record terminator ($) following the cross-referenced segment declaration (preceding the cross-referenced field declarations).

FILENAME = COMPFILE, SUFFIX = FOC, $
 SEGNAME = COMPSEG, SEGTYPE = S1, $
   FIELD = COMPANY,   ALIAS = CPY,     FORMAT = A40,           $
 SEGNAME = DIVSEG,  PARENT = COMPSEG, SEGTYPE = S1, $
   FIELD = DIVISION,  ALIAS = DV,      FORMAT = A20,           $
   FIELD = DIVMGR,    ALIAS = NAME,    FORMAT = A12,           $
 SEGNAME = ADSEG,   PARENT = DIVSEG,  SEGTYPE = KU,
  CRSEGNAME = IDSEG, CRKEY = DIVMGR, CRFILE = PERSFILE,
   FIELD = NAME,      ALIAS = FNAME,   FORMAT = A12, INDEX = I,$
   FIELD = DADDRESS,  ALIAS = ADDRESS, FORMAT = A24,           $
   FIELD = DDOB,      ALIAS = DOB,     FORMAT = YMD,           $
 SEGNAME = PRODSEG, PARENT = COMPSEG, SEGTYPE = S1, $
   FIELD = PRODUCT,   ALIAS = PDT,     FORMAT = A8,            $
   FIELD = PRODMGR,   ALIAS = NAME,    FORMAT = A12,           $
 SEGNAME = BDSEG,   PARENT = PRODSEG, SEGTYPE = KU,
  CRSEGNAME = IDSEG, CRKEY = PRODMGR, CRFILE = PERSFILE,
   FIELD = NAME,      ALIAS = FNAME,   FORMAT = A12, INDEX = I,$
   FIELD = PADDRESS,  ALIAS = ADDRESS, FORMAT = A24,           $
   FIELD = PDOB,      ALIAS = DOB,     FORMAT = YMD,           $

DIVMGR and PRODMGR are described as CRKEYs. The common alias, NAME, is automatically matched to the field name NAME in the PERSFILE data source. In addition, the field declarations following the join information rename the ADDRESS and DOB fields to be referred to separately in reports. The actual field names in PERSFILE are supplied as aliases.

Note that the NAME field cannot be renamed since it is the common join field. It must be included in the declaration along with the fields being renamed, as it is described in the cross-referenced data source. That it cannot be renamed is not a problem, since its ALIAS can be renamed and the field does not need to be used in reports. Because it is the join field, it contains exactly the same information as the DIVMGR and PRODMGR fields.

The following conventions must be observed:


Top of page

x
Joining From Several Segments in Several Host Data Sources: Multiple Parents

At some point, you may need to join to a cross-referenced segment from two different host data sources at the same time. If you were to describe a structure like this as a single data source, you would have to have two parents for the same segment, which is invalid. You can, however, describe the information in separate data sources, using joins to achieve a similar effect.

Consider an application that keeps track of customer orders for parts, warehouse inventory of parts, and general part information. If this were described as a single data source, it would be structured as follows:

You can join several data sources to create this structure. For example:

The CUSTOMER and ORDER segments are in the ORDERS data source, the WAREHOUSE and STOCK segments are in the INVENTRY data source, and the PRODINFO segment is stored in the PRODUCTS data source. Both the INVENTRY and ORDERS data sources have one-to-one joins to the PRODUCTS data source. In the INVENTRY data source, STOCK is the host segment. In the ORDERS data source, ORDER is the host segment.

In addition, there is a one-to-many join from the STOCK segment in the INVENTRY data source to the ORDER segment in the ORDERS data source, and a reciprocal one-to-many join from the ORDER segment in the ORDERS data source to the STOCK segment in the INVENTRY data source.

The joins among these three data sources can be viewed from the perspectives of both host data sources, approximating the multiple-parent structure described earlier.


Top of page

x
Recursive Reuse of a Segment

In rare cases, a data source may cross-reference itself. Consider the case of a data source of products, each with a list of parts that compose the product, where a part may itself be a product and have subparts. Schematically, this would appear as:

A description for this case, shown for two levels of subparts, is:

See the Creating Reports manual for more information on recursive joins.


Information Builders