Cullen Programming

JAVA Pipelines

Design Notes, Known Problems and Limitations

Design Notes

This implementation of PIPELINES will attempt to follow closely to the John Hartmann's original IBM implementation and documentation of CMS Pipelines.

Currently no more that one secondary input/output is supported for all filters that use secondary I/O.

Some features of CMS/TSO Pipelines have meaning only on EBCDIC machines and those features are not implemented here.

Some features of CMS/TSO Pipelines are IBM z/VM and z/OS platform specific and are not implemented here.

A DeadLock Monitor is currently in development.

General Limitations

Default storage allocation is 64M bytes for product execution. This should accommodate at least 50 piped stages. The value is user modifiable and can be easily adjusted higher.

JAVA Pipelines can currently accommodate up to 100 stages.

The piped stream buffer size will currently accommodate up to 1 million characters.

Each input record must be delimited by either a newline character (ASCII X'0A') or a carriage return and linefeed (ASCII X'0D0A').

Each PIPE can have up to 99 STEM inputs but is limited to a single STEM output.

The SORT stage can accommodate 10,000 80 byte records using the default storage allocation.

Known Problems

Some syntax checking and error information messages remain to be developed.

[Return to Index]

[Return to Cullen Programming Home Page]