Adding Value before loading data into a Cube

[ARNOJ]

Morning,

I have a view in TM1 that I have created with my TurboProcess and that I need to put in a Cube.

My view is reading attribute from an account to convert it into an analytical account. Therefore some account have exactly the same analytical account

Everything is fine, until I have check the data

I Have use CellPutN : (Value, Account,Entity,Month,Year....)

However when two lines are strickly the same in my view I will have only 1 value.

By example
AccountA EntityA Jan Y2013 : 100
AccountA EntityA Jan Y2013 : 10

In my destination cube, I will have only 100 instead of having 110.

How I can sum the value ?

Thanks you

Regards

[jim wood]

Have you tried creating a process using the GUI? Try setting a process to accumulate and see what it does.

[ARNOJ]

The GUI ?

I don't have access to mapping.

[java_to_tm1]

Have you tried simply picking up the existing value before storing the new one?

existing_val = CellGetN (Account,Entity,Month,Year....)

CellPutN ( existing_val + Value, Account,Entity,Month,Year....)

[Alan Kirk]

The GUI ?

I don't have access to mapping.

Thank the gods for that. You're better off staying away from that robotically constructed cr@p and learning real coding anyway.

As java_to_tm1 said you need to aggregate the value, but since you're on 10.1 you can also take advantage of the CellIncrementN function, one of the rare improvements these days which aren't just marketing "chrome" but are of substantial practical value in running a TM1 system efficiently. Given that it exists, grab it with both hands and use it instead of the CellPutN function.

With regard to your inability to access the Map tab, this suggests that you don't currently have access to the data source from your machine. (Presumably you create the view dynamically, and it doesn't exist when you're editing the process.) If you create the view manually before opening the process to edit it you should be able to access (or, better still, ignore) the mapping tab

[java_to_tm1]

since you're on 10.1 you can also take advantage of the CellIncrementN function, one of the rare improvements these days

Ever since those ghastly performance killers called "Link cubes: auto-generate feeders" come up in 10.1, I've learnt to be wary of "new improved" functionality.

Has anyone done the performance comparison of CellIncrementN and (CellGetN + CellPutN) ? Let me test that out and put the results up.

[Alan Kirk]

since you're on 10.1 you can also take advantage of the CellIncrementN function, one of the rare improvements these days

Ever since those ghastly performance killers called "Link cubes: auto-generate feeders" come up in 10.1, I've learnt to be wary of "new improved" functionality.

There's a vast difference between "anything that is automatically generated out of a Java application", and "a new TI function". The former involves some developers who seem utterly disconnected from real world usage making a guess about what the user intends to do based on huge over-generalisations, the latter (in this case) is probably nothing more than a wrapper around exactly the same functionality that previously took three lines of code.

Has anyone done the performance comparison of CellIncrementN and (CellGetN + CellPutN) ? Let me test that out and put the results up.

By all means do. If you can find more than a billionth of a second difference in the speed and can actually measure it, I'll be suitably surprised. I'm prepared to be quite blunt and scathing when Iboglix get it wrong but they'd need to have really made a conscientious effort to screw this one up; it's just an overdue piece of practical functionality that someone thankfully finally noticed the absence of and did something about. Unlike the Java apps it's nothing actually new, just a more efficient implementation of what was there.

[ARNOJ]

Thanks you all for your answer

I have tried the both solution and it was still not working

I have decided to export my view in a TXT file format. Therefore I realized that the view I was using was wrong. However after I have made a pivot table based on my TXT File i can see that both where working

the real issue was the construction of my view and not how to add value

After solving my vie, the both solution had been working properly.

Regards

[jim wood]

Thank the gods for that. You're better off staying away from that robotically constructed cr@p and learning real coding anyway.

While I agree with you, using the GUI to generate code when your very new to TI is good way to pick up basic ways of doing things. For any new user, for me it's good way to see the basics of what's going on.

[ARNOJ]

I have been told that IBM will not support any TurboIntegrator created in Perspectives and then run in Performance Modeler.

IF you do so (run a Turbo Integrator created under Perspectives, in Performance Modeler) no support from IBM and they will ask you to redo your process directly into Performance Modeler

Anyone heard the same thing ?

[jim wood]

I've not heard that one. With the migration to 10.2 and with advent of Cafe I wouldn't be surprised though.

[Duncan P]

Performance Modeler does not run the TI. The TI is run in the server and as far as I am aware the server has only one kind of process, regardless of where it was created.

[Alan Kirk]

Thank the gods for that. You're better off staying away from that robotically constructed cr@p and learning real coding anyway.

While I agree with you, using the GUI to generate code when your very new to TI is good way to pick up basic ways of doing things. For any new user, for me it's good way to see the basics of what's going on.

If the mapping tab was intuitive, I'd agree with you. But IMHO it isn't; the post that we had here on one occasion asking what "ASLS" is being but one case in point. You therefore need to invest time in learning the thing, and my feeling is that that time could be better utilised working your way through the Reference Guide anyway. It's sometimes said that no knowledge is ever wasted; my feeling is that one exception to that is "learning an unintuitive and esoteric GUI that you'll get no long term value from and will dump as soon as you know what you're doing".

I have been told that IBM will not support any TurboIntegrator created in Perspectives and then run in Performance Modeler.

IF you do so (run a Turbo Integrator created under Perspectives, in Performance Modeler) no support from IBM and they will ask you to redo your process directly into Performance Modeler

Anyone heard the same thing ?

I haven't heard that either. At face value it doesn't make a lot of sense; there is no actual difference between the TI language in either environment. There may be a couple of different values in the .pro file indicating which environment the thing came from (I haven't checked) but the actual language syntax is identical. The only difference that I can recall coming across in 10.1 didn't even relate to processes but to chores; when creating a chore in the Editorsaurus you could set the repeat cycle up to 999 days. In Muddler it was only 365.

However it's possible that what you heard might be a reference to code that was generated in Perspectives out of the Map tab, since obviously the automation process (such as it is) will not necessarily be the same between the Editorsaurus and Performance Muddler. (I have to admit that I didn't look at many of the automation features in Muddler and don't know what, if anything, it has similar to the Map tab. Or if I once did, I don't recall it.) For hand crafted code it shouldn't make any difference where it was created.