Hello everyone,

I've run into an issue with PAfE (latest release; 2.0.69.3) and a Cube with UNDEFVALS defined.
We are using Office 365:
In Architect and in the CubeViewer the data is shown correctly:

As expected, with UNDEFVALS the cube differentiates between 0 and NULL.
But in PAfE (Quick Report in this case, but also in Dynamic and Custom Report) it only shows 0 for all cells:


There is no numberformating on the measure.
I've tried different Excel-Numberformats.
I've made sure this Excelsetting is set:


Without this Excelsetting everything is blank:

So it seems that PAfE handles all values the same, even though there should be a difference.

Does anyone have any idea how to make it work in PAfE or if this is a known issue?

Thanks and BR,
Matthias

------------------------------
Matthias Mazaj
------------------------------
#PlanningAnalyticswithWatson

Did this work before the upgrade?
What version of the TM1 Server?

You may want to log a case with support for the PAFE team to investigate the issue.

------------------------------
paul YOUNG
------------------------------

It did/does work in Perspectives but not in PAfE. The version doesn't seem to matter, as I've tested it with a few different ones.

For the test above I used TM1 Server 11.8.9, PAW 2.0.69, PAfE 2.0.69.

It does seem like a bug to me, so I plan to log a case, but I just wanted to check if someone else had this issue, in case I missed something or did something wrong.

Thanks for your response

------------------------------
Matthias Mazaj
------------------------------

Original Message:
Sent: Fri November 19, 2021 04:56 AM
From: paul YOUNG
Subject: UNDEFVALS in PAfE

Did this work before the upgrade?
What version of the TM1 Server?

You may want to log a case with support for the PAFE team to investigate the issue.

------------------------------
paul YOUNG
------------------------------

HI Matthias,
Not sure if you saw Ryan's thread but may be of interest.

------------------------------
George Tonkin
------------------------------

Original Message:
Sent: Fri November 19, 2021 05:04 AM
From: Matthias Mazaj
Subject: UNDEFVALS in PAfE

It did/does work in Perspectives but not in PAfE. The version doesn't seem to matter, as I've tested it with a few different ones.

For the test above I used TM1 Server 11.8.9, PAW 2.0.69, PAfE 2.0.69.

It does seem like a bug to me, so I plan to log a case, but I just wanted to check if someone else had this issue, in case I missed something or did something wrong.

Thanks for your response

------------------------------
Matthias Mazaj
------------------------------


Original Message:
Sent: Fri November 19, 2021 04:56 AM
From: paul YOUNG
Subject: UNDEFVALS in PAfE

Did this work before the upgrade?
What version of the TM1 Server?

You may want to log a case with support for the PAFE team to investigate the issue.

------------------------------
paul YOUNG

Hi,

I've been through this issue with IBM already on support tickets and have two RFE open. Different use cases, but I think it is the same general problem.

There seems to have a been a design decision taken early on, perhaps brought over from CAfE to suppress "error" messages from the view of users, when using TM1 forumla (DBRW / DBSW)
#N/A is not visible - So you can't see circular or divide by zero errors. ( https://ibm-data-and-ai.ideas.ibm.com/ideas/PAOC-I-745 )
Undefvals not showing.
DBS formula evaluating to the cell contents if a send is not possible, rather than throwing an error. (RFE here https://ibm-data-and-ai.ideas.ibm.com/ideas/PAOC-I-331)

Whilst I can see how you might make this design call, it is massively restrictive, cuts out many use cases and in the case of the DBS issue creates lots of work to check your DBS is executing properly.

(It might not be a design decision, it could be that the method used to implement the forumla can only handle a numeric type and therefore isn't capable of display error strings or empty cells.)

My #NA use case was slightly different in that a customer was using a ruled 1 divided by 0  #N/A error to provide an input mask to users in for budget entry at the N level.  An unusual use case to be sure, but one that has worked well for them in the legacy world for a numbers of years.

In the support ticket (TS005241710), we went around in circles about if this was a defect or not.  Basically the product is working as designed, so its not a defect.

Hopefully not to much of a side track from your issue.

If I can beg everyones indulgence to vote for the RFEs listed above that would be great. Could I also encourage everyone to get more involved in the RFE process by voting on any that catch their eye. It's an important communication channel into IBM.

Steve Rowe
Technical Director
www.infocat.co.uk

------------------------------
Steven Rowe
------------------------------

Hello Steven,

thanks for the reply and the links to the RFE.
The first RFE is somewhat related to UNDEFVALS as I've tried using a Rule with "['X'] = N: UNDEF;" to create a "#N/A" in the cell (which is correctly shown in cubeviewer). In Excel it's an empty cell, instead as a zero (where as the NULL and 0 cells show as 0).


​But since I want to differentiate between NO INPUT and "0" as an input, i cannot apply this rule to the cell.

I highly doubt this is working as intended, since this works in Perspectives and not in PAfE.

Perspectives (Active Form):



------------------------------
Matthias Mazaj
------------------------------

Original Message:
Sent: Mon November 22, 2021 06:25 AM
From: Steven Rowe
Subject: UNDEFVALS in PAfE

Hi,

I've been through this issue with IBM already on support tickets and have two RFE open. Different use cases, but I think it is the same general problem.

There seems to have a been a design decision taken early on, perhaps brought over from CAfE to suppress "error" messages from the view of users, when using TM1 forumla (DBRW / DBSW)
#N/A is not visible - So you can't see circular or divide by zero errors. ( https://ibm-data-and-ai.ideas.ibm.com/ideas/PAOC-I-745 )
Undefvals not showing.
DBS formula evaluating to the cell contents if a send is not possible, rather than throwing an error. (RFE here https://ibm-data-and-ai.ideas.ibm.com/ideas/PAOC-I-331)

Whilst I can see how you might make this design call, it is massively restrictive, cuts out many use cases and in the case of the DBS issue creates lots of work to check your DBS is executing properly.

(It might not be a design decision, it could be that the method used to implement the forumla can only handle a numeric type and therefore isn't capable of display error strings or empty cells.)

My #NA use case was slightly different in that a customer was using a ruled 1 divided by 0  #N/A error to provide an input mask to users in for budget entry at the N level.  An unusual use case to be sure, but one that has worked well for them in the legacy world for a numbers of years.

In the support ticket (TS005241710), we went around in circles about if this was a defect or not.  Basically the product is working as designed, so its not a defect.

Hopefully not to much of a side track from your issue.

If I can beg everyones indulgence to vote for the RFEs listed above that would be great. Could I also encourage everyone to get more involved in the RFE process by voting on any that catch their eye. It's an important communication channel into IBM.

Steve Rowe
Technical Director
www.infocat.co.uk

------------------------------
Steven Rowe
------------------------------

It's very frustrating indeed. All of these "design decisions" are driving us away from pafe.

------------------------------
Ryan Clapp
------------------------------

Original Message:
Sent: Mon November 22, 2021 06:25 AM
From: Steven Rowe
Subject: UNDEFVALS in PAfE

Hi,

I've been through this issue with IBM already on support tickets and have two RFE open. Different use cases, but I think it is the same general problem.

There seems to have a been a design decision taken early on, perhaps brought over from CAfE to suppress "error" messages from the view of users, when using TM1 forumla (DBRW / DBSW)
#N/A is not visible - So you can't see circular or divide by zero errors. ( https://ibm-data-and-ai.ideas.ibm.com/ideas/PAOC-I-745 )
Undefvals not showing.
DBS formula evaluating to the cell contents if a send is not possible, rather than throwing an error. (RFE here https://ibm-data-and-ai.ideas.ibm.com/ideas/PAOC-I-331)

Whilst I can see how you might make this design call, it is massively restrictive, cuts out many use cases and in the case of the DBS issue creates lots of work to check your DBS is executing properly.

(It might not be a design decision, it could be that the method used to implement the forumla can only handle a numeric type and therefore isn't capable of display error strings or empty cells.)

My #NA use case was slightly different in that a customer was using a ruled 1 divided by 0  #N/A error to provide an input mask to users in for budget entry at the N level.  An unusual use case to be sure, but one that has worked well for them in the legacy world for a numbers of years.

In the support ticket (TS005241710), we went around in circles about if this was a defect or not.  Basically the product is working as designed, so its not a defect.

Hopefully not to much of a side track from your issue.

If I can beg everyones indulgence to vote for the RFEs listed above that would be great. Could I also encourage everyone to get more involved in the RFE process by voting on any that catch their eye. It's an important communication channel into IBM.

Steve Rowe
Technical Director
www.infocat.co.uk

------------------------------
Steven Rowe
------------------------------