Chapter 14: Advanced Data Set Information

This chapter expands on the information presented in the chapter Files and Data Sets. It covers the following subjects:

Data set organizations
Record formats
Logical record length
Mapping to local file formats
Dynamic PDS member lists
Support for Generation Data Groups

Data Set Organizations

The data set organizations supported by MSS are shown in the following table.

Abbreviation	Organization	Comment
GDG	Generation data group
IS	Indexed sequential
ISU	Indexed sequential, unmoveable
PDSM	Partitioned data set member	Micro Focus organization
PO	Partitioned
POU	Partitioned, unmoveable
PS	Physical sequential
PSU	Physical sequential, unmoveable
VSAM	Virtual storage access method

Note:

Although the direct access (DA) organization is not supported explicitly, you can emulate it by using the VSAM organization and relative record (RR) record format.

Record Formats

The following table shows the record formats for all the data set organizations except VSAM.

Abbreviation	Format	Control Characters	Record Delimiters	Comment
F	Fixed	None	None
FA	Fixed	ANSI	None
FB	Fixed blocked	None	None	Same as F format
FBA	Fixed blocked	ANSI	None	Same as FA format
FBM	Fixed blocked	Machine	None	Same as FM format
FBS	Fixed blocked spanned	None	None	Same as F format
FM	Fixed	Machine	None
FS	Fixed spanned	None	None	Same as F format
LSEQ	Line sequential	None	Carriage return/line feed
U	Undefined			Results unpredictable
V	Variable	None	Micro Focus type
VA	Variable	ANSI	Micro Focus type
VB	Variable blocked	None	Micro Focus type	Same as V format
VBA	Variable blocked	ANSI	Micro Focus type	Same as VA format
VBM	Variable blocked	Machine	Micro Focus type	Same as VM format
VBS	Variable blocked spanned	None	Micro Focus type	Same as V format
VM	Variable	Machine	Micro Focus type
VS	Variable spanned	None	Micro Focus type	Same as V format

Note: Blocked and spanned record formats are accepted by MSS but have no meaning in a PC environment.

The following table shows the record formats for VSAM files.

Abbreviation	Format	Control Characters	Comment
ES	Entry sequence	None
KS	Key sequence	None
LS	Linear sequence	None	Not supported
RR	Relative record	None

Notes:

A cataloged VSAM file has two extra properties, LRECL Avg (average record length) and LRECL Max (maximum record length). If these two properties have the same value, MSS assumes that the file has fixed length records and does not use record delimiters to separate the records. If they have different values, MSS assumes that the record has variable length records and uses the Micro Focus type delimiters to separate records.
If you have a program compiled to use EBCDIC, but that program reads a file with a RECFM of LSEQ and a character set of ASCII, the file's data is automatically converted to EBCDIC as part of the read. This automatic conversion is useful when reading files that do not contain hexadecimal data, such as control cards.
You allocate VSAM data sets to the MSS catalog using New on the ESMAC catalog page.
If your application uses record-level locking on shared VSAM files, you might find that files larger than 1 GB (gigabyte) cannot be accessed. This is because the default record locking mechanism in MSS uses the top two bits of the file address for locking and compression bits, thus reducing the address space for the maximum offset from 4 GB to 1 GB for a shared idxformat 3 or 4 file. If you encounter this problem, set the LOCKTYPE Compiler directive to 2 when you compile the program affected. With this setting, the file handling mechanism creates a separate file for storing record locks, with the same basename as the file and a .lck extension.

Logical Record Length

MSS works with logical record lengths as defined in the following table:

Record Format	Logical Record Length
Fixed	Actual record length
Variable	Actual record length + 4
LSEQ	Maximum record length

Note: If you have a program compiled to use EBCDIC, but that program reads a file with a RECFM of LSEQ, LRECL of 0 and a character set of ASCII, MSS automatically converts the file's data to EBCDIC as part of the read. This automatic conversion is useful when reading files such as control cards, that do not contain hexadecimal data.

Mapping to Physical File Formats

MSS writes data sets to physical files on the PC using the following organization and record format mapping:

Mainframe		Physical File
Data Set Organization	Record Format	File Organization	Record Format	Comment
DA, DAU, PS, PSU	Any fixed (Fnn) or variable (Vnn) format	Sequential	Micro Focus
DA, DAU, PS, PSU	LSEQ	Line sequential	Micro Focus	Always uses ANSI character set. Can only be edited with a text editor
GDG				Catalog entry only
IS, ISU	Any fixed (Fnn) or variable (Vnn) format	Indexed sequential	Micro Focus
PDSM	Any fixed (Fnn) or variable (Vnn) format	Sequential	Micro Focus
PDSM	LSEQ	Line sequential	Micro Focus
PO, POU				Catalog entry only
VSAM	ES	Sequential	ESDS
VSAM	ES	Indexed sequential	ESDS	Alternate index defined
VSAM	KS	Indexed sequential	Micro Focus
VSAM	LS	Not supported
VSAM	RR	Relative	Micro Focus

Support for Generation Data Groups

A generation data group (GDG) consists of two parts: a GDG model and a series of physical sequential (PS) data sets, called biases (also known as generation data sets) that are associated with the GDG.

Creating Generation Data Groups

There are two main ways to set up a GDG:

By clicking Add on the ESMAC Catalog page. To define the model, specify the data set organization DSORG as "GDG". To define a bias, specify the DSORG as "PS" and provide the fully-qualified data set name. For example, if the model has the data set name MAG.TEST, the first bias might be named MAG.TEST.G0001V00, the second MAG.TEST.G0002V00, and so on (gaps in the numbering sequence are allowed).
By running an MFJAMS job that defines the GDG model and biases.

The following example illustrates a job that can be used to create a generation data group. STEP00 creates the GDG model and STEP01 creates five new GDG biases.

//MFIDMFGD JOB 'MFIDMFGD',CLASS=A,MSGCLASS=A,NOTIFY=MFIDF
//STEP00 EXEC PGM=IDCAMS
//SYSOUT DD SYSOUT=*
//SYSPRINT DD SYSOUT=*
//AMSDUMP DD SYSOUT=*
//SYSIN DD *
DELETE MFIDMF.TEST.GDG
DEFINE GENERATIONDATAGROUP(NAME(MFIDMF.TEST.GDG) 
       EMPTY NOSCRATCH LIMIT(96))/*
//STEP01 EXEC PGM=IEFBR14
//SYSUXX DD DSN=MFIDMF.TEST.GDG(+1),DISP=(NEW,CATLG),SPACE=(TRK,(5,2))
//SYSUXY DD DSN=MFIDMF.TEST.GDG(+2),DISP=(NEW,CATLG),SPACE=(TRK,(5,2))
//SYSUXZ DD DSN=MFIDMF.TEST.GDG(+3),DISP=(NEW,CATLG),SPACE=(TRK,(5,2))
//SYSUXA DD DSN=MFIDMF.TEST.GDG(+4),DISP=(NEW,CATLG),SPACE=(TRK,(5,2))
//SYSUXB DD DSN=MFIDMF.TEST.GDG(+5),DISP=(NEW,CATLG),SPACE=(TRK,(5,2))

Five biases named MFIDMF.TEST.GDG.G0001V00, MFIDMF.TEST.GDG.G0002V00, MFIDMF.TEST.GDG.G0003V00, MFIDMF.TEST.GDG.G0004V00 and MFIDMF.TEST.GDG.G0005V00 are created by STEP01.

A bias can be processed by its relative number, its actual reference or as a part of the entire generation data group. For example, each of the following statements could be executed after running the previous job.

//SYSUT1 DD DSN=MFIDMF.TEST.GDG,DISP=SHR

This would cause the entire generation data group MFIDMF.TEST.GDG, consisting of a set of concatenated data sets MFIDMF.TEST.GDG.G0001V00 through MFIDMF.TEST.GDG.G0005V00, to be referenced.

//SYSUT1 DD DSN=MFIDMF.TEST.GDG(0),DISP=SHR

The relative number 0 would cause the data set named MFIDMF.TEST.GDG.G0005V00 to be referenced.

//SYSUT1 DD DSN=MFIDMF.TEST.GDG(-1),DISP=SHR

The relative number -1 would cause the data set named MFIDMF.TEST.GDG.G0004V00 to be referenced.

//SYSUT1 DD DSN=MFIDMF.TEST.GDG(+1),DISP=SHR

The relative number +1 would cause a new data set named MFIDMF.TEST.GDG.G0006V00 to be referenced.

//SYSUT1 DD DSN=MFIDMF.TEST.GDG.G0003V00,DISP=SHR

The actual reference G0003V00 would cause the data set named MFIDMF.TEST.GDG.G0003V00 to be referenced.

The increment of numbers occurs per job, so if you had two consecutive steps that referenced +1, then each step would be using the same data set.

Generation Data Group Restrictions

This section describes restrictions when using GDGs in MSS:

The maximum number of host GDG biases that you can reference is 99.
MSS does not support versioning of the GDG biases, so the only biases that versioning works for are those with a data set name ending with V00. For example, if the GDG model name is MFIDMF.TEST.GDG then the GDG biases must have the form MFIDMF.TEST.GDG.G0001V00 ... MFIDMF.TEST.G9999V00, with gaps in the numbering sequence. If you attempt to use a bias that ends, for example, in V01, it will not be found when GDG names are resolved during program execution.
A GDG model and all of its generation data sets (GDSs) must reside on the same platform. It is not possible to have a mixture of GDSs where some reside on the mainframe and others locally.
If you are using relative bias numbers to reference GDSs in your JCL, for example MFIDMF.TEST.GDG(-4) or MFIDMF.TEST.GDG(+1), the following restrictions apply:
- If the GDSs are located on a Mainframe Access server, you cannot reference any numbers greater than zero. The only permitted numbers are 0 through -n, where n is the maximum number of GDSs supported by the GDG model.
- If the GDSs are located on a PC or network, all processing is supported with only the limitations that are already documented in this section.
EMPTY is not supported.
A further restriction applies when you are allocating G0000V00 data sets using the G0000V00 numbers EXPLICITY rather than using positive relative generations; for example, where you are using:
```
//NEWFILE DD DSN=TEST.GDG.G0002V00,DISP=(NEW,CATLG)
```
rather than:
```
//NEWFILE DD DSN=TEST.GDG(+1)
```
For each GDG, the system catalog stores the number of the G0000V00 data set that corresponds to the current generation of that GDG. Where G0000V00 data sets are allocated using the G0000V00 numbers explicitly, you must set the current generation on the model to 0. This forces the catalog API to recalculate the correct current G0000V00 number by finding the number of the highest G0000V00 data set currently cataloged for the model.

You can reset the current generation number by DELETING and REDEFINING the model without uncataloging the G0000V00 data sets. You can do this using IDCAMS, as illustrated below:
```
     DELETE TEST.GDG  
     DEFINE GENERATIONDATAGROUP     - 
       (NAME(TEST.GDG)              - 
       SCRATCH                      - 
       LIMIT(02)) 
```
For example, assume that we have a cataloged GDG called TEST.GDG, which currently has two generations corresponding to TEST.G0001V00 and G0002V.00. If a data set TEST.GDG.G0007V00 is allocated directly, the current generation field in the GDG model must be reset. This will force the GDG processing to scan all the associated G0000V00 data sets to recalculate the correct value for the current generation, which in this case would be G0007V00. Failure to do this would cause a JCL error when the normal GDG cycle process attempts to allocate TEST.GDG.G0007V00 and finds that the data set already exists.