TM1 Forum

Hi all, im trying to write a feeder, and i kinda need to know if is if posible to use dimix function on feeders.

i have 2 cubes, and in the second one i have the same dimension as the cube1 but with another name, but the exact same elements.
I'm not very familiar with the function dimix, but i saw some generated rule with performance modeler and they do something like this to validate the elements

in cube1 i have DimensionX
and cube2 i have DimensionZ that is the same as DImensionX but with a diferent name.


['Measue']= N: if(dimix('DimensionX',!DimensionZ)<>0, DB('Cube2',!DimensionZ,!Dim1,!Dim2,!Dim3),CONTINUE);

with this i can validate the elements , because they have the same index as the cube 1.


in Cubo2 i have some multiplication, and that data is coming from Cube1, then I want to make a feeder for cubo1, but I have the problem that the name of the dimension is different, so I want to use dimix
i do not know if you understand me, hope you get the idea..sorry for my bad english tho.

what i did was this:

['Measure']=> if(DIMIX('DimensionX,!DimensionZ)<>0,DB('Cube2',!DimensionZ,!Dim1,!Dim2,!Dim3),CONTINUE);

but with no good results.

if someone has any experecien with this, pls let me know, any tip will be appreciated
Thanks

You don't need to use dimix. When writing rules or feeders across cubes, dimension elements which match will link and unmatched elements will fail silently. Note you can't use CONTINUE in a feeder.
I'd say, therefore would work fine.

It's hard to help because you haven't specified in which cubes these actual rule statements are housed. It's like your first statement would be in Cube 1 because you are trying to pull in values from Cube2, however, your rule is using the !DimensionZ syntax a couple of times and that is not valid in Cube1 because DimensionZ is not in that cube. Everything seems to be mish-mashed together and I can't tell what's going on.

That being said, why do you think you even need to use DIMIX? If the two dimensions are identical then just use the name. If an element or two doesn't match then it will be skipped, much like a JOIN in SQL.

Rule in Cube 1: Rule in Cube 2:

Tom's code snippet should do the trick.

But I've always wondered about the need for creating a copy of a dim with exactly the same elements:

Kazêh wrote: i have the same dimension as the cube1 but with another name, but the exact same elements.

I can understand there's a valid case for doing this if you have an "approval" type application built on that dim (You would need a copy of the dim for cubes that shouldnt have approval hierarchy).

I've also seen cases where there are two (or more) time dimensions (with possible different levels of granularity).

Just curious if there are any typical cases where you'd need an exact copy of a dim.

(different thread perhaps?)

Kazêh wrote: what i did was this:

['Measure']=> if(DIMIX('DimensionX,!DimensionZ)<>0,DB('Cube2',!DimensionZ,!Dim1,!Dim2,!Dim3),CONTINUE);

but with no good results.

Although I agree with the other posters, I will let you know which is the right code for a conditional feeder:
It is not very intuitive, but once you understand how it works, you will remind it the next time

Sebastian.Klein wrote:Although I agree with the other posters, I will let you know which is the right code for a conditional feeder:

Yes, but the point here is that you do not need a conditional feeder. If you have two dimensions A and B, in different cubes, and they have overlapping members then doing this:
Then only those members in DimB that match an element in DimA will get fed. It already is conditional, there is no need to do a DIMIX check and then an IF statement.

Thanks for all the answers.

Normally do not post the original code because I feel that if I did, it would be like asking you guys to solve my problem, and what I want are tips and try to learn to do it alone. also there are people who get angry if you ask them...and they send me to read the manual again, and come when am a litle wiser...so!:

As someone says, I can not use the same dimension because I'm using it as a dimension of hierarchy in different applications.
The real rule is a litle more complex, i will post it right away:
im really not good at writin complex feeders, i dont know if this is complex or not, but i always try to follow the example ['B']= ['A']*['C']; ['A']=> ['B'] , ['A']=> ['C]
and separate the calculations into blocks and see if I can build something

A* (B(1)/B(2))* C

I'm basically multiplying a value from cube "Primas UPA RRC", then a division between a date and a few days of validity and then multiply it by a percentage, that's on simple words.
im still trying to pull this out, i will take your tips and let you know as soon as posible if it's working or not.


Thank you all for your support

i did it guys, i write a feeder in cube Calculo RRC UPA part 1 like this:

['Fecha Aux']=>DB('Calculo RRC UPA Part 2',!Canal,'Devengacion',!Gerencias Tecnicas RRC UPA,!Ramo,'TOTAL',!Mes Emision,!Metricas RRC);

and i worked perfectly. Thanks for all your tiips, sorry the trouble.

thread Closed