Re: Skipping lots of releases - Was: Looking at Power 7

On 1/24/11 12:44 PM, rob@xxxxxxxxx wrote:

Yes I do encourage people to stay current. And, technically I can
easily imagine corruption occurring following this method. However,
I find it hard to argue with experience. And it seems that these two
gents have a bit of it. Granted, there's the possibility of doing so
many of these that you tend to not remember minor irritants during an
upgrade that you already have workarounds for.

Since for many years I supported the database feature and even provided install\upgrade\PTF mentoring and design review for components utilizing the database in products and features, I have experience in both the theory and in practice.

There would generally be no APARs for wanton [mis]utilization of unsupported\undocumented install methods; the effects could be so wildly unpredictable that attempting to document them is near futile and certainly anathema. Certainly nobody wants to incur the cost of IBM having to predict what some symptoms might be for installing 7.1 over v4r3 product and pseudo-system libraries and directories.

While many "conversions" are little more than a CL request to "CPYF OldVersion NewVersion FMTOPT(*MAP *DROP)" which alone is probably mostly "safe", consider that there may be additional work to effect "complete" conversions. One example is SYSPROCS which was [on v4r3 I believe] a physical file, which on v4r4 converts to a VIEW and the physical file is since SYSROUTINES, and the IS_DETERMINISTIC changed from an inverse, both by column name and thus the stored value. Later that file had IASP details added for which some post-install conversion likely performed by the OS DB feature to update that data, but that support would only be triggered [even exist in code] on releases up to N+2 since the data was added.

Regards, Chuck