Re: search & compile options

This same technique is covered in this IBM Redbook on Modernizing in Chapter 9
Modernizing IBM i Applications from the Database up to the User Interface and Everything in Between
www.redbooks.ibm.com/redbooks/pdfs/sg248185.pdf

Easiest way if you don’t have a CM package managing this for you.

Thanks
Jeff Tickner
Arcad Software

date: Mon, 26 Oct 2015 09:22:18 -0400
from: Charles Wilt <charles.wilt@xxxxxxxxx>
subject: Re: [WDSCI-L] search & compile options

The easy way is to have a CMS that track the dependencies and can
automatically recompile everything for you.

I have one, and have used it successfully in the past to do that very thing.

However, now-a-days, I think the easiest way is not to recompile any
programs at all.

No, I don't mean LVLCHK(*NO).

What I do is the following, assume I have a PF called MYPF.

1) I create a new PF, well SQL table actually, named MYPF_T
2) I copy the data in MYPF to MYPF_T
3) I repoint all existing LFs to MYPF_T
4) I convert MYPF from a PF to a LF over MYPF_T

In steps 3 & 4, the key is to ensure that the LF's have an explicit format
(ie. the previously existing fields are specified) and that the new field
is left out.

Done correctly, all previously existing objects will have the same record
format ID as they had before, so no programs need be recompiled.

Going forward, make sure all RLA is done through a LF with an explicit
format and you can add any fields you need with a simple ALTER TABLE / CHGPF

You'll need new LF(s) that contain the new field(s) for any programs that
need access. Don't let this worry you. Assuming that the new LF has the
same key(s) as an existing, it adds no overhead besides a few bytes on
disk.

The technique, logical file access only, has been around awhile and the
process has most recently been outlined as the way to move from DDS to DDL.

You'll mind more discussion on the Midrange and RPG lists.

Charles