Re: Modernization and multi-member files

Wilt, Charles wrote:

So?

If you look at the DB2 SQL manual for LUW and i, you'll see the following in both:

Can the application see uncommitted changes made by other application processes?
NC - YES (DB2 for i only)
UR - Yes
CS - No
RS - No
RR - No

Can "updated" rows be read by other application processes that are running at an isolation level other
than UR and NC?
NC - Yes (DB2 for i only)
UR - No
CS - No
RS - No
RR - No

Now if you want to say that to many i jobs run at NC or UR, then I'd agree to that.

And that once again is the beauty of the platform. You can write applications that do not need commitment control and you can have visibility of "uncommitted" changes (which is common sense if you think about it, since *all* changes are uncommitted when you don't use commitment control). Whatever your position on commitment control, you can design your architecture as best fits your needs with DB2. Others don't give you that flexibility.

Can flexibility lead to bad decisions? Yes. But as a programmer I still like being able to make those decisions as I need to. Remove those decisions, and I'm no longer an analyst, or an architect, I'm a coder. Is it bad to be a coder? No. But I didn't spend 30 years to become one.

Joe