We have a bunch of VSAM files on our mainframe that
house our critical print production data, such
as Inventory, Bill Of Materials, Project Management System,
etc.

In each of the files, there are "sub-tables", ie:
repeating OCCURS or OCCURS DEPENDING fields.

There are many REDEFINES. The REDEFINES might map
4 binary half-word integers over the same space as
a text field or a date field or 3 PIC 9 fields or (...).

Fields within the front of the record tell you how
to treat / read / parse the rear of the record.

If you foolishly read a binary integer, but is REALLY
is random text, and you are a COBOL programmer, your
program ABENDS.

This makes it kind of tedious to track down bad data.

We have about 20 years of this type of data in these
files.

Previously when I needed to work with this data, I
would need a MF programmer to extract it, or I'd dump
it myself and then parse it out, a bit at a time.

Very tedious and error prone. And I could NEVER
update it. I would always need a MF programmer for
that, on a case by case basis.

So we have installed software called tcAccess that lets
me have an ODBC connection to these MF files. I can
read/write them with select and insert/update statements.

All within Perl on Linux.

It works.

I'm amazed.

So I get to code new applications on the Linux boxen
running Perl, while integrating with the MF data without
the concept of staged file transfers.

I'll create some clearly defined interfaces to funnel all
activity through. Which will then allow me to use this
type of data from another system as well.