ATTRS/N or DB}ElementAttributes for rule?

[tosca1978]

Hi all,

I was just wondering what the best practice is for referencing an attribute from within a rule. For example I could write the following 2 pieces of code within a rule to achieve the same result:

Code: Select all

ATTRS('Time_Day',!Time_Day,'Act For')@='Actual',

or

Code: Select all

DB('}ElementAttributes_Time_Day',!Time_Day,'Act For')@='Actual',

But I was wondering if either one performed better than the other or if there was a good reason to use one over the other. In worksheets I tend to use the DBRW approach into the }ElementAttributes cube rather than a DBRA formula if I'm copying the formula down a lot of rows becuase I think this performs better.

Cheers

[Alan Kirk]

I was just wondering what the best practice is for referencing an attribute from within a rule. For example I could write the following 2 pieces of code within a rule to achieve the same result:

Code: Select all

ATTRS('Time_Day',!Time_Day,'Act For')@='Actual',

or

Code: Select all

DB('}ElementAttributes_Time_Day',!Time_Day,'Act For')@='Actual',

But I was wondering if either one performed better than the other or if there was a good reason to use one over the other. In worksheets I tend to use the DBRW approach into the }ElementAttributes cube rather than a DBRA formula if I'm copying the formula down a lot of rows becuase I think this performs better.

Interesting question. I recall Andy Key stating that he had been told that a DBR worksheet function is faster than a DBRA, but whether the same is true of rules functions I couldn't say; it would be interesting to know if his informant provided some details on that. (Whether the difference would be material is another matter.)

[ioscat]

don't really know if it is a good test:
1) created dim2 with 10^x elements

Code: Select all

vCount = 1;
vDimSize = 100000;
WHILE (vCount<vDimSize);
vCount_s = str(vCount, 10, 0);
dimensionelementinsert('dim2','', vCount_s,'n');
attrputs(vCount_s, 'dim2', vCount_s, 'test');
vCount = vCount + 1;
END;

cube1 rule

Code: Select all

skipcheck;

['a'] = N: NUMBR(ATTRS('dim2',!dim2,'test'))
+NUMBR(ATTRS('dim2',!dim2,'test'))
+NUMBR(ATTRS('dim2',!dim2,'test'))
+NUMBR(ATTRS('dim2',!dim2,'test'))
+NUMBR(ATTRS('dim2',!dim2,'test'))
+NUMBR(ATTRS('dim2',!dim2,'test'))
+NUMBR(ATTRS('dim2',!dim2,'test'));

cube2 -

Code: Select all

skipcheck;

['a']=N:NUMBR(DB('}ElementAttributes_dim2',!dim2,'test'))
+NUMBR(DB('}ElementAttributes_dim2',!dim2,'test'))
+NUMBR(DB('}ElementAttributes_dim2',!dim2,'test'))
+NUMBR(DB('}ElementAttributes_dim2',!dim2,'test'))
+NUMBR(DB('}ElementAttributes_dim2',!dim2,'test'))
+NUMBR(DB('}ElementAttributes_dim2',!dim2,'test'))
+NUMBR(DB('}ElementAttributes_dim2',!dim2,'test'));

didn't get any difference

[Andy Key]

Interesting question. I recall Andy Key stating that he had been told that a DBR worksheet function is faster than a DBRA, but whether the same is true of rules functions I couldn't say; it would be interesting to know if his informant provided some details on that. (Whether the difference would be material is another matter.)

Unfortunately not. We were only talking about worksheet functions.