problem with text data in a cube (Cognos TM1. 10.2)

[Wirt]

Hello everyone!
I have a problem with text data in a cube; in the rules has been active function FEEDSTRINGS, but not correctly inserted into cell text data, I insert the data array, for example, 15 cells, but is inserted only 5 or 10, and in different cells. I disabled the function FEEDSTRINGS in the rules. All became fine inserted. But recently started having problems again: in the text cell changing any value, I click recalculate, and begins a long time to recalculate.
For example, I change the date in the text cell, and the recalculation has taken more than 3000 seconds, then I just overload the server, because in the operating console I can not disable the process.
If someone had a similar problem, I'll be grateful for the help.

[gtonkin]

HI Wirt,
Not exactly sure what you are doing but sounds like you have cells in which you have text i.e. strings - this information is captured or populated but *not* rule-derived.
If the cells you are waiting for to be recalculated are based on strings, strings are never cached and always get recalculated from scratch.
If not rule derived, you do not need FEEDSTRINGS unless you are using these values to feed other cells. You would rarely use FEEDSTRINGS unless you need to feed from strings.

Please provide more detail to allow people to understand and assist better.

[Wirt]

Thanks, gtonkin, for the answer.

The question is - whether it is necessary feed text values? Perhaps the reason that the rule saves long, that I feed text values?
the following example:
Dimension has elements -
1. manual correction ( through a Pick List, select Manual or payment )
2. Calculated element
rule:

FEEDSTRINGS;
SKIPCHECK;

['Calculated element'] = N: IF ( DB ('Cub',!Dim1,!Dim2,'manual correction')@='manual', STET, ['other condition'] );

FEEDERS;

['other condition']=>['Calculated element'];
['manual correction']=>['Calculated element'];

Another example:
Dimension has elements -
1. Date (text elem)
2. Period
3. Calculated element
Rule:

FEEDSTRINGS;
SKIPCHECK;

['Calculated element'] = N: IF (
YEAR ( DB ('Cub',!Dim1,!Dim2,'Date') = YEAR ( !Period )
&
MONTH ( DB ('Cub',!Dim1,!Dim2,'Date') = MONTH ( !Period ) , ['calculation 1'], ['calculation 2'] );

FEEDERS;

['Date']=>['Calculated element'];
['calculation 1']=>['Calculated element'];
['calculation 2']=>['Calculated element'];

I write correctly? or text elements in feeders are not involved?

Thanks.

[gtonkin]

From the information you have posted, 'Manual Correction' is the text element which is input i.e. not calculated, no need to feed it.
If the condition is manual then you N: rule will STET - no need to feed this either
If the condition is not manual then you are deriving the value for 'calculated element' - this would need to be fed, generally from the dependent variable i.e ['other condition']=>['Calculated element'];
I cannot see that you ever need to feed from 'Manual Correction' for any values as 'Other Condition' is already feeding.

Similarly with your 'Date' feeder - either 'Calculation 1' or 'Calculation 2' should be used but possibly not Date, especially if it is a string.
In this case, as you are probably going to end up with a value for either the True or False part of the IF statement, you could choose to feed from a N level cell that you know will always be populated. If Date in the example is a string, based on the fact that you are comparing it with a value returned by the YEAR() function, you could use this. All depends on the dimensionality of the two cubes as you probably do not want to feed to C: levels by default.

[Wirt]

Thanks for the answer, remove feeders for text cells.

[tstroumbos]

Hey guys, I briefly read this, and want to see if it relates to an issue I'm seeing:

We have FEEDSTRINGS because we do have some rules which are triggered by the entry of text. However, it seems when we input into ANY string cell, it now refires all the feeders, which is a substantial process. Does that seem correct or is there something else in play here?

EDITING: also, as we're entering text at a consolidated level, in TM1 Top it's mostly stuck on "cubecellspreadviewarray"....why would it be trying to spread a text value?

I didn't build this model, just coming in to support, so maybe there's something poorly done that I'm not aware of.


Thanks!

[tstroumbos]

I answered my own question. Removing FEEDSTRINGS enabled use of strings at consolidated levels without a re-firing of the entire rule file. That seems like a bug...

[BrianL]

However, it seems when we input into ANY string cell, it now refires all the feeders, which is a substantial process. Does that seem correct or is there something else in play here?

Changing any string cell value should only process feeders that fire from that single cell. It should not reprocess feeders for every cell in the cube.

EDITING: also, as we're entering text at a consolidated level, in TM1 Top it's mostly stuck on "cubecellspreadviewarray"....why would it be trying to spread a text value?

That's the API architect/perspectives uses for pretty much all cell writes. This lets it use the same API call for single cell writes, spreads, and even implicit spreads to consolidations.