TM1 Forum

From http://wiki.olapforums.com/index.php?title=View:

TM1 does not distinguish between views which are created for data analysis purposes, and those which are intended as a Data Source for TurboIntegrator. There is a school of thought that Cognos should create such a distinction and separate data source views from analytical views, given that data source views are seldom suitable for viewing and are often too large to be displayed. There has been little if any indication that Cognos is inclined to address this issue, however.

My suggestion for resolving this would be related to another enhancement post I'm about to make: security on views. If you could set views to only be opened by certain groups, then you could limit TI views to administrators only.

The other option would be to create something that is not a "view" in the current meaning of the term. For want of a better term, the name "TISource" would work. This could be created in a similar fashion to the way we create views for TI processes now (right click on cube, export as ascii data, build, save) but could not be opened in the cube viewer by a user, only by a TI process.

Any other ways of handling this problem that we can suggest to Cognos?

Curiously, the ancients know better. In olden days of v6.0 queries and views were two different things. lumping them together was a poor idea. Browser views and process queries may work the same way, but they are so not the same thing.

/signed (in blood)

Have consistently stated to Applix / Cognos (including development managers/etc) that this is my biggest TM1 bugbear. I've quite a few but this is the one that I think is the most urgent to sort - turn it into a .TIS (Turbo Integrator Source) or somesuch, not a public view.

It's point blank insanity to have it be something a user can view.

First of all apologies for the grammatical mistakes I may do, I am French...

The way I go around this:
1 - I create a TI to build up the view (public). In the prolog I create subsets for each dimension belonging to my view using a while statement (looping on the index) or you could use dynamic subset - then depending on what the view needs to be I insert the relevant function and then insert each element into the subset (based on the criterias in the loop).
2 - I then create a TI using the above view, run my data in, and delete the subsets and view in the epilog.
3 - I create a TI which executes both processes

Doing it this way goes around the issue of having a public view used by a TI and available to end-users.

Nick SAINT-MARTIN

Hi Nick,

I tend to do the same thing as well but the TI to create a view from scratch and populate all the subsets is pretty long winded. If you could have "system queries" then if would be a lot easier to deal with.

Cheers,

...and unfortunately it's not an option on 9.0 because there is a serious bug that corrupts any view created by TI that has more than 2 dimensions. The same bug exists in 9.1 up to a certain patch, but not sure which one, and Cognos support said Engineering are "unable" to create a patch for 9.0 to fix it in that version...

The TM1 powers that be have advised me that they're aware of the need for separate views/TI data sources, and also the need for sharing views between non-admin users, but that these additions are not expected in any currently-planned release.

At least they're on their radar I guess.

Martin

Not entirely reassuring when we've been telling them for at least 3 years tho!