I think you've missed my point.
Fixpaks are supposed to be a vehicle for delivering fixes for problems. By
altering it to a vehicle for adding new functionality, you increase the risk
that - by adding some new features - you introduce new problems. In
particular, 8.1.4 (as per your post requesting feedback thereon) did just
that - it broke basic query processing that worked previously (PMR
10781,211,848. Hotfix 9272).
Accordingly my feedback is that you would be best off leaving fixpaks as a
means of delivering fixes and not as a means of enhancing functionality.
This is especially so given the frequency of fixpaks and the consequent
limited testing new features receive and consequent large risk of
introducing new defects.
As a user, I am used to performing exhaustive testing with new product
versions. I do not expect to have to do this for fixpaks, in particular as
the standard initial PMR response is "install the latest fixpak and see if
that fixes your problem".
We stumbled over this bug "by accident" - it was only by luck that we
weren't in production. The DB2 team wanted to go live with 8.1.4 (based
mostly on our significant 8.1.3 testing), and I was the one who insisted on
more testing time. Although outnumbered, it was finally accepted that if
there was a problem, I was going to be the one yelled at by the users, so
that if I wanted more testing time, I could have it. And what I was worried
about was actually a spurious SQLCode -911. I wasn't expecting to find that
simple queries gave bad results.
So not only do you need more testing (we agree on that), but you should stop
using FIXpaks as a means of adding new features. Leave them to do what they
say the do - bundles FIXes.
"Serge Rielau" <sr*****@ca.e ye-bee-m.com> wrote in message
news:bs******** **@hanover.toro lab.ibm.com...
Not much I can say about that. It's a well known problem that testing
can never we enough.
--
Serge Rielau
DB2 SQL Compiler Development
IBM Toronto Lab