Alias not unique (TI)

[flesh2011]

Hi.

Need help with a TI process.

case:

Dim1 generic names. --> Alias name as is ===> TI to change alias to new alias names

Element1 2010 2011
Element2 2011 2012
Element3 2012 2013
Element4 2013 2014

......ok, example?

Using a version cube to set all the dimension variables/parameters.
When i try to roll the TI, i get attributes not unique.

Is there any way to get this to work without restarting server?

thanks in advance

[jim wood]

Attribute changes are not applied until the end of the tab. You might want to remove the attribute in the prolog and then rebuild it. Then insert the new values in Data / Epilog. That way you guarantee that the cache for the alias is clear.

[flesh2011]

Hi.

Tried what you said, but cant get it to work.
When i try to manualy edit in the element attributes i get the message in the attatchment.

what does that mean?
Can`t see that i have duplicates ?

Thanks

[AmbPin]

Hello,
Are you trying to change the value of the alias?
If so then you may be being caught out by an alias that is still in use by another element. Consider the following simple dimension:-

Element - Alias
Year1 - 2012
Year2 - 2013

If I wanted to increment one year forward then you might try this in the attributes editor:-
Year1 - 2013
Year2 - 2014

However this would fail because until all of the changes are committed 2013 is still associated with Year2. If this is what is affecting you, then you will first need to clear all of the attribute values before attempting the update.

[jim wood]

I have seen problems with aliases being cached. When you did as I said, did you destroy the alias? Try destroying it then complete a data save and rebuild it. I would however check very carefully to see if you have any alias / element names that are the same as what you are trying change before taking such a drastic step. If you still don't get any joy, try recycling the service which will for sure clear any cache,

Jim.

[qpf]

Hi,

I have a very similar Problem with an alias attribute. It's a dimension with SAP Fiscal and Booking Periods. Each Element called fiscper consists of the number of the year and the number of the month with a zero between, i.e 2014003 for March 2014.
Each element is in several Hierarchies. Now I want an alias called buper for each N-Element, the only difference ist that the zero between year and month is missing : 2014003 ==> alias 201403
I get the same errors as mentioned above in TI-Process or interactive in the attribute Editor. (The Attribute is not unique. Value 201403 can not be set as an attribute for element 2014003 in dimension SAP_Fiscper_V_Alias. )
I saved befor changing attributes, and I restarted TM1 for another trial. I always got the error.

What else could be wrong ?

Thanks

Dirk

[Wim Gielis]

Hello

When doing this manually, it's working fine. TM1 10.2.

Create a new dimension, insert the element 2014003 in the Dimension Editor.
Go to the editor to change the attributes: add an alias, call it "test", and set 201403 as the value for the alias.

Works fine.

Can you limit the data source to make sure you're not running into a different problem, then the problem specified here?
And use a test dimension temporarily, to see whether it works fine in a completely new empty dimension.

[jim wood]

Dirk,

TM1 treats zeros and spaces as nulls so in your case TM1 thinks both of them are the same value. You'll need to make them different by ading an under score or something like that,

Jim.

[Michel Zijlema]

Dirk,

TM1 treats zeros and spaces as nulls so in your case TM1 thinks both of them are the same value. You'll need to make them different by ading an under score or something like that,

Jim.

Hi Jim,

I'm not sure whether I'm misinterpreting your remark, but TM1 is certainly not treating zeroes as nulls. So 2014003 and 201403 are absolutely two different things...
(see also Wim's testing results above).

Michel

[qpf]

Hi,

in a new Dimension, it's possible to create the alias 210403 for the element 2014003.
Then I checked my TI-Process and I found an error, the elements 2014003 and 201403 were created. So the alias-Error was correct.
Then I fixed the bug, but it's still not possible to create the aliases. Then I looked for the Element 2014003 in the TI-Script generated Dimension in the Subset Editor.
If I filter with 2014003, The left window of the subset editor, shows the Element eleven times( its member of in the year hierarchy( 1x) , and member of the
Cumulation hierarchies March to December 2014, the Alias-element 201403 was not found ), so the eleven found Elements seems to be correct. But I only exspected to find the Element one time.
How can this behaviour be declared ? I think that each element has to be unique in one Dimension.
Then I tried to set alias manuylly in the Dialog. Now it was possible ! So I tried it with TI.
The code for creating the Elements positioned in DATA, code for aliases positioned in METADATA. This doesn't work, although no errors were reported. After creating a second process only
for the creating of the aliases, everything runs fine. All aliases were correctly created.
So the treatment of zeros seems to be changed, so that only spaces will further on make problems.

The the only remaining problem are the eleven found elements.

Thanks

Dirk.

[Michel Zijlema]

If I filter with 2014003, The left window of the subset editor, shows the Element eleven times( its member of in the year hierarchy( 1x) , and member of the
Cumulation hierarchies March to December 2014, the Alias-element 201403 was not found ), so the eleven found Elements seems to be correct. But I only exspected to find the Element one time.
How can this behaviour be declared ? I think that each element has to be unique in one Dimension.

I'm sure the 2014003 element will be unique in you dimension. If you click the All button in the subset editor and do a wildcard search/filter on 2014003 you will see the element only once. Also if you select the 11 occurencies you see in your subset and look in on the right side (properties pane enabled), you will see all 11 occurences have the same dimension index number.

The code for creating the Elements positioned in DATA, code for aliases positioned in METADATA. This doesn't work, although no errors were reported. After creating a second process only
for the creating of the aliases, everything runs fine. All aliases were correctly created.

Here's where you're going wrong. Metadata (dimension structures) need to be updated in the METADATA tab, not DATA. If you update the dimension structure on the METADATA tab, you can load the aliases on the DATA tab in the same process...

Michel

[qpf]

alternatively placing the Code into Prolog and Epilog works also fine. I think the Reason ist that TI saves the Dimension after a Tab is finished, and the elements can be found only after saving.

Dirk

[jim wood]

Dirk,

TM1 treats zeros and spaces as nulls so in your case TM1 thinks both of them are the same value. You'll need to make them different by ading an under score or something like that,

Jim.

Hi Jim,

I'm not sure whether I'm misinterpreting your remark, but TM1 is certainly not treating zeroes as nulls. So 2014003 and 201403 are absolutely two different things...
(see also Wim's testing results above).

Michel

Yes, I got myself mixed up a little on that one. Thinking about another problem while answering another. Thanks for pointing it out. Hopefully I didn't confuse anybody too much.