Possibilities for Endusers

[danielclp]

Hi all!

I just read through some topics and saw that it is possible for endusers to define elements in dimension through a front-end.
http://www.tm1forum.com/viewtopic.php?p=15472
My question now is if we talk about a dimension with postions or measures if it is possible for endusers to connect these elements with some easy calculation operators through a front-end.

For example:
1. Add: position A
2. Add: position B
3. Add: position C
4. Define: position A * position B = position C

[jim wood]

Multiplication is normally handled by rules rather than weighting. Are you trying to get the users to defined rules? If so you to use (with baited breathe) Performance Modeller. It's the only place I know of that they'll be able to do this without a security risk,

Jim.

[danielclp]

The idea is to allow them define only the calculation operators *, +, - and / through a frontend with unlimited positions.

[Duncan P]

Does it have to be instant or could they be required to run a TI to make their calculations into rules?

[Alan Kirk]

Multiplication is normally handled by rules rather than weighting. Are you trying to get the users to defined rules? If so you to use (with baited breathe) Performance Modeller. It's the only place I know of that they'll be able to do this without a security risk,

If you're waiting for one of my trademark "I just so luuuuurve Muddler" rants, I'm afraid that you're in for a disappointment... but it is worth mentioning page 2 of the Muddler manual:

To create and manage Cognos TM1 Applications, you must be a member of the ADMIN group on the TM1 server.

Insight is for users, Muddlers is for administrators. More specifically for Admins who want to put their sanity at risk. And of course if you're in the Admin group it's open for you to modify rules using the more conventional methodologies.

That having been said... I'd be inclined to be looking at the design of the cube to start off with. There shouldn't be any need to be adding rules "on the fly" to a cube, certainly not in the sort of instance that danielclp is describing. For that I'd suggest putting a "measures" type dimension in which contains a generic calculation based on the other elements. In that case the new elements could be added to the other dimensions and the calculations would already apply. (Not least because the issue of feeders probably needs to be addressed.)

If that absolutely, positively cannot be done, then I think Duncan's solution of using TI is going to be the only viable one, albeit a potentially rather fiddly one.

[danielclp]

Thanks for all your answers.
But I want only a hint if this is possible and I think all your posts give me the answer that it is possible.

[tomok]

Thanks for all your answers.
But I want only a hint if this is possible and I think all your posts give me the answer that it is possible.

Actually, it's not possible.

[Alan Kirk]

Thanks for all your answers.
But I want only a hint if this is possible and I think all your posts give me the answer that it is possible.

Actually, it's not possible.

Yeah it is; the way Duncan alluded to. Create a front end to a TI to add the elements, have the TI write the corresponding rules and feeders to a text file, and use RuleLoadFromFile to suck that file back in.

It is, however, probably the definitive example of the sometimes-yawning chasm between what is technically possible, and what is actually a good idea.

(With the possible exception of my propeller-headed friends who used Java to kinda-sorta automate Excel as I alluded to in a recent post. That one was probably marginally worse.)

[tomok]

Thanks for all your answers.
But I want only a hint if this is possible and I think all your posts give me the answer that it is possible.

Actually, it's not possible.

Yeah it is; the way Duncan alluded to. Create a front end to a TI to add the elements, have the TI write the corresponding rules and feeders to a text file, and use RuleLoadFromFile to suck that file back in.

It is, however, probably the definitive example of the sometimes-yawning chasm between what is technically possible, and what is actually a good idea.

(With the possible exception of my propeller-headed friends who used Java to kinda-sorta automate Excel as I alluded to in a recent post. That one was probably marginally worse.)

I generally give answers that include some qualitative background in it. When I said it was not possible, I meant using the native features of TM1, as well as a reasonable work-around. There is no way in Hades that anyone would be able to pull that one off except for maybe Duncan himself, lotsaram, or some other EXTREMELY experienced developer and even then, would not really function in the way that the OP is envisioning. That's why I said it's not possible.

[jim wood]

Tomok you have a point. You get most software to do anything. You can get excel to do most of things that TM1 can do, but would you really want to. But as the guy says he has answers and options.

[Alan Kirk]

Actually, it's not possible.

Yeah it is; the way Duncan alluded to. Create a front end to a TI to add the elements, have the TI write the corresponding rules and feeders to a text file, and use RuleLoadFromFile to suck that file back in.

It is, however, probably the definitive example of the sometimes-yawning chasm between what is technically possible, and what is actually a good idea.

I generally give answers that include some qualitative background in it. When I said it was not possible, I meant using the native features of TM1, as well as a reasonable work-around.

A curious distinction, given that all of the functionality items that I mentioned are native features of TM1.

There is no way in Hades that anyone would be able to pull that one off except for maybe Duncan himself, lotsaram, or some other EXTREMELY experienced developer

I suspect that rather more of us could do it, but would choose not to.

and even then, would not really function in the way that the OP is envisioning.

Given that the OP gave only a vague outline of what he was envisaging, I'd be reluctant to jump to that conclusion.

The more general principle that monkeying with rules on the fly in an operational environment is not a recommended practice... that one is a safe conclusion.