System 1032

Back to Basics, Part 5—Introducing Attribute Formats
By Tym Stegner

Display Format Compared to Storage Format
Contrary to a common assumption, the format of an attribute does not describe how the attribute data is physically stored within a dataset. System 1032 stores data values in the most efficient manner possible, which usually you cannot actually display. Storage is based on factors from attribute qualities such as RANGE and PRECISION.
This article begins a discussion on how attributes values are displayed when you extract or report them. We examine in some depth the FORMAT quality of an attribute definition. The FORMAT quality is a string of encoded characters that controls how that attribute data value is displayed by the WRITE or PRINT commands.

Default Attribute Formats by Data Type
The data type itself governs the default display. In System 1032, the G format uses the default format for the data type. Any attribute value can be displayed using the general formats. You can override the default formats by specifying an alternate display using the FORMAT quality in the ATTRIBUTE command. You can also override a defined display format with a WRITE or PRINT command, if you want to use a different format temporarily.

Table 1 lists the default formats for each data type.

Table 1. System 1032 default display formats

Although the following example uses variables, instead of attributes, the default formats are the same, because like attributes the defaults are based on a data type.

1032> variable EXPIRATION date initially '12/31/2007'
1032> variable UNIT_PRICE decimal initially 12.31
1032> variable UNITS integer initially 1000
1032> show var * format
  Variable EXPIRATION Date Format D3
  Variable UNIT_PRICE Decimal Format N10.2
  Variable UNITS Integer Format N10
1032> print expiration unit_price units
EXPIRATION    UNIT_PRICE       UNITS
----------  --------------  -----------
12/31/2007           12.31         1000
1032>

Using the General Formats
The next levels of display control you can exercise are the general numeric display, the N format, or the general text display, the A format. (The A format and the other text formats are discussed in detail in a subsequent article.)

You can use the N syntax of Nwidth.decimal to control the display of digits when the data type is Decimal, Integer, or Real. Note that the width segment of the N format describes the total width of the display field, including the decimal point.

1032> variable sample decimal init 1234.56
1032> show variable sample format
Variable SAMPLE Decimal Format N10.2
1032> print sample sample format n15.2 sample format n6.1
    SAMPLE            SAMPLE          SAMPLE
--------------  -------------------  ---------
       1234.56              1234.56     1234.6
1032>

Numeric Formats
The Integer, Real, and Decimal data types have many options for display. Beyond the common integer and floating point formats, System 1032 supports exponential formats for scientific notation, as well as hexadecimal and octal renderings.

Logical Formats
The data type Logical represents True, False or MISSING values. You can display them as text, the Lwidth format, or in binary, the Bwidth format.

Binary display might be used for scientific reporting, often written as B1. The Bwidth format syntax displays True/False values as 1 or 0 (zero), in the specified width of field. If a width greater than 1 is specified, the single numeric digit is preceded by the requisite number of spaces.

The text logical format uses the Lwidth syntax. Similar to the B format, the width of the L format controls the spacing to the left of the value of T or F by filling in with spaces.

1032> variable YesNo logical init false
1032> print yesno yesno fmt L5 yesno fmt B1
YESNO  YESNO  YESNO
-----  -----  -----
  F        F    0

Date-related Formats
System 1032 is very comprehensive when it comes to displaying date and/or time data values. The following Date_Time format syntax looks incomprehensible until we pick it apart in Table 2.

D{[W][d[YY][.t[M]]}

Table 2. Identifying the parts of Date_Time format syntax

Table 3. D format details for date and date-time data

In the following example, a birthday date is displayed using two formats.

1032> variable dayBirth date_time initially ‘11/28/1960 19:35’
1032> print dayBirth format dW dayBirth format d1
DAYBIRTH        DAYBIRTH
---------  ------------------
   Monday  November  28, 1960
1032>

In a similar fashion, the Time and Time_Span data types, shown in Table 4, use enumerated formats for output of elapsed and clock time. The Time and Time_span formats are rendered using the TSspecifier.width syntax.

Table 4. TS format for time and time-span data

DEFINE STREAM CCAJRNL   WITH SCOPE=SYSTEM -
 PARALLEL=(CCAJPERM,CCAJGDG) MINAVAIL=2
DEFINE DATASET CCAJPERM WITH SCOPE=SYSTEM OLD -
       DSN=YOUR.PERM.CCAJRNL
DEFINE DATASET CCAJGDG  WITH SCOPE=SYSTEM -
 CATALOG GEN=+1 CYL PRI 100 UNIT=DISK204 
 DSN=YOUR.GDG.CCAJRNL

You will create two identical copies of CCAJRNL.

Both datasets must have exactly the same DCB and space allocation parameters. Also ensure that each dataset is allocated to a different volume and channel for better performance.

If the write operation to either dataset fails due to an I/O error or a dataset full condition, Online terminates since MINAVAIL=2. If this happens, you must run recovery to reestablish the Online environment.

The Figure 2 schematic illustrates the process of writing to a parallel stream. The Production Online has five servers, 50,000 buffers, and eight CCAJRNL buffers. The parallel journals, CCAJPERM and CCAJGDG, are shown on two separate disks.

Figure 2. Writing a parallel stream

//CCARF DD DSN=YOUR.PERM.CCAJRNL,DISP=SHR

In Summary
If your organization archives CCAJRNL, consider defining CCAJRNL as a parallel stream and create the archive copy during the Online run. This saves what is usually an extra copy step after Online termination and may even improve system availability, since the Online bounce will not have to wait for the copy step.