I really like your train of thought here.
It will work perfectly, the only issue is some of the files I am doing this with contain 10 million records. This may take awhile.
I was hoping to do something where I didn't have to copy the data twice - once to create a journal record, and a second time to do the ETL.
If I could drop the cust record format into the JOESD field using an SQL join to create a Cartesian join of all rows to all other rows. But maybe that is copying data twice too.
I will give this a try - thanks Mark!
-----Original Message-----
From: midrange-l-bounces@xxxxxxxxxxxx [mailto:midrange-l-bounces@xxxxxxxxxxxx] On Behalf Of Mark S Waterbury
Sent: Monday, October 01, 2012 10:38 AM
To: Midrange Systems Technical Discussion
Subject: Re: make cust file look like a journal record - format
Joel:
Why go to all the trouble of trying to make a file that "looks just like
a journal record"?
Why not just use the real system journal facilities and then you can use
DSPJRN, as you do now?
For example, you could do something like this:
1. create a temporary library:
CRTLIB LIB(TEMPLIB)
2. create a temporary journal and journal receiver:
CRTJRNRCV JRNRCV(TEMPLIB/TEMPJRN0001)
CRTRJN JRN(TEMPLIB/TEMPJRN) JRNRCV(TEMPLIB/TEMPJRN0001)
3. create a copy of the file, but with no data, in the temporary library:
CRTDUPOBJ OBJ(filename) FROMLIB(filelib) OBJTYPE(*FILE)
TOLIB(TEMPLIB) DATA(*NO)
4. start journaling the temporary copy of the file:
STRJRNPF FILE(TEMPLIB/filename) JRN(TEMPLIB/TEMPJRN) IMAGES(*AFTER)
OMTJRNE(*OPNCLO)
5. copy all the trecords from the original file to the temporary copy,
to get them journaled:
CPYF FROMFILE(filelib/filename) TOFILE(TEMPLIB/filename) MBROPT(*ADD)
Then you can just issue DSPJRN against TEMPLIB/TEMPJRN and send the
output to an *OUTFILE, as I presume you are doing now ......
When all done, you can issue:
DLTLIB TEMPLIB
Does that help?
All the best,
Mark S. Waterbury
On 10/1/2012 11:19 AM, Stone, Joel wrote:
I have pgmA which reads cust journal changes and sends them off for an ETL app.
I would also like to be able to send the entire cust file using the same pgmA with no changes.
How can I make the cust record format look just like a journal record (the format that DSPJRN creates).
For example:
Cust file format:
CustName 30
CustAddr 30
CustCity 20
-----
Total 80
I want this to look like a DSPJRN record so I can use the same pgmA to load up all CUST records for an initial load.
Desired format:
Field Field
Text Name Type Length
Length of entry JOENTL S 5 ---------------------|
Sequence number JOSEQN S 10 |
Journal Code JOCODE A 1 |
Entry Type JOENTT A 2 |
Date of entry: Job date f JODATE A 6 |
Time of entry: hour/minut JOTIME S 6 |---------- journal header fields
Name of Job JOJOB A 10 |
Name of User JOUSER A 10 |
Number of Job JONBR S 6 |
Name of Program JOPGM A 10 |
Name of Object JOOBJ A 10 ----------------------|
.
.
.
Entry Specific Data - Var JOESD A 80 <--------------------------------- I want my CUST data here!!
Most of the journal fields can empty - but I want my CUST data to be in JOESD. Can SQL or OPNQRYF do this?
I think I need to take fields JOENTL thru JOOBJ and initialize those. Then take all CUST fields CUST.* and string them together (alpha, packed, zoned) into the one JOESD field.
Is this possible without using an HLL?
Thanks!
______________________________________________________________________
This outbound email has been scanned for all viruses by the MessageLabs Skyscan service.
For more information please visit http://www.symanteccloud.com
______________________________________________________________________
