I usually don’t do ‘release’ posts, but I think this one is significant for a couple of reasons. 2.9 is an official long-term support release, which means that IBM will provide only fixes to this 2.9.x branch, but no new functionality will be incorporated. None of the previous 2.x releases had such designation.

IBM even managed to demonstrate why this is important. To cite the Hyrum’s law

With a sufficient number of users of an API, it does not matter what you promise in the contract:

all observable behaviors of your system will be depended on by somebody.

so when IBM decided to change the way security is evaluated for MDX subsets (from “first run MDX, then apply security” to “apply security and then run MDX”), surely enough, things broke for somebody in an upgrade project. The fact that this change was considered so ‘intuitive’ and ‘trivial’ that it wasn’t announced in change log was an extra twist that makes Quebit’s journey so painful. But you had to do certain things with your user interfaces to get the issue, we haven’t seen this in our upgrades (converting dynamic subsets to static for the win).

And the beauty of the ‘long-term support’ release is that you’d get only fixes, not new features like this :) To IBM’s credit, they’ll revert this subset behaviour in the next IF for 2.0.9.1.

Secondly, 2.0.9.1 has a massive fix list (some of those are ours) and I’m starting to feel that the core server parts in this release are ‘10.2.2 fp7’-grade solid and it wouldn’t give us a lot of trouble in the implementations. It’s a feeling that you crave as a consultant, you have enough challenges without wrestling with software. All this is about Tm1 server + Tm1web though, PAW (and by extension PafE) is whole different kettle of fish.