Comments on: RPG Program Files in Update Mode

By: The Most-Watched IT Questions: February 8, 2011 - ITKE Community Blog

The Most-Watched IT Questions: February 8, 2011 - ITKE Community Blog — Tue, 08 Feb 2011 06:50:06 +0000

[...] 8. CharlieBrowne and TomLiotta team up again to help Sangay with a question regarding RPG program files in update mode. [...]

By: tomliotta

tomliotta — Fri, 28 Jan 2011 21:54:16 +0000

ODP’s will not be shared between these two programs unless it is done in CL…

That depends on how the file is created. Usually, the default on the create is SHARE(*NO), but that can be changed when the command is run or by running CHGPF SHARE(*YES) or CHGLF SHARE(*YES).

If the already has the SHARE(*YES) attribute, then you’d need to run an override to avoid sharing when it’s unwanted.

Objects should be created with the attributes they should have normally. Overrides should only be used for exceptions.

Tom

By: sangay

sangay — Fri, 28 Jan 2011 05:48:33 +0000

how Force write ratio will help?

Thanks

By: charliebrowne

charliebrowne — Fri, 28 Jan 2011 05:42:57 +0000

You can change the Force Write Ratio to 1

By: sangay

sangay — Fri, 28 Jan 2011 03:55:23 +0000

Ye thats what i am also thinking about. P2 will recieve the old record since the ODP’s will not be shared between these two programs unless it is done in CL which calls P1.

By: tomliotta

tomliotta — Thu, 27 Jan 2011 21:59:07 +0000

P1 updates a particular record (say using CHAIN). Now P2 also want to update the same record (Say using CHAIN). will P2 recieve a new data or old data from File FILE1?

As long as P1 has completed the update operation, P2 should get the updated record.

I’ve never tried, but it might be possible to cause P2 to get the old record by ensuring separate ODPs, calling P2 before a COMMIT, overriding for sequential blocking so that the record CHAINed to in P2 was already in the P2 buffer due to a prior positioning and possibly other deliberate code structures; but I’d expect it to take almost deliberate actions.

If P1 called P2 before completing the update, then P2 should run into a record lock condition.

Tom