15.10 - Row-count - Call-Level Interface Version 2

Teradata Call-Level Interface Version 2 Reference for Mainframe-Attached Systems

prodname
Call-Level Interface Version 2
vrm_release
15.10
category
Programming Reference
featnum
B035-2417-035K

Row‑count is a four byte unsigned integer field that specifies the maximum number of rows to return in the response buffer. A value of zero imposes no row limit. When specified before a request is initiated, subject to the control of the Change‑option option, it limits the rows returned in each response buffer until the value is changed. When specified in the DBCAREA before a Fetch function the Change‑options option is not relevant, and the value will limit the rows in a response buffer obtained by that Fetch. If a Fetch function returns an unfetched parcel from an existing response buffer, no new response buffer is obtained and the DBCAREA Row‑count value has no effect. To achieve the desired effect, an application changing Row‑count before a Fetch must understand the relationship of parcels to buffers. A response buffer may contain multiple parcels; all parcels in a buffer are returned by Fetch one at a time before the next response buffer of parcels is requested from the Database; if the Two‑response‑buffers option is 'Y', the default, then the next response buffer is filled before it is needed. Since exactly when such buffer prefilling is done is not under application control, a Two‑response‑buffers option of 'Y' is not recommended when a using non‑zero Row‑count.

Row‑count does not affect the results of an SQL request, only the packaging of the results in response buffers.

 

In this language...

The variable name for Row‑count is...

COBOL

ROW‑COUNT

PL/I

ROW_COUNT

C

rowCount

IBM Assembler

DBCIRCNT

 

This routine...

Does this for Row‑count...

DBCHINI

writes

DBCHCL

reads (RSUP; FET; IRQ; IWPF)

 

Row‑count is used by...

To...

applications

write

The default value is 0.

When Row‑count is used with multi-statement requests, the rows from different statements are never combined into one response buffer. So the last response buffer for each statement may contain fewer rows than requested.

When a non-zero Row‑count is specified, the Two‑response‑buffers option should probably be set to 'N' because 'Y' is used to maximize fetching performance while Row‑count is used to reduce it.