Planning Analytics

Hi,
If I create a view with MDX and then assign it as a datasource to a view, this works well but...

If the MDX query returns no results then (as far as I can tell) I run into errors that I can not trap or suppress.

ViewCreateByMDX(sCubeName , sObjectName, sMDX, booTempOn);

#Check that the view got created (failed attempt to trap error)
If(ViewExists(sCUbeName , sObjectName)=1);

#Make the view the source
     DatasourceCubeview=sObjectName;

Else;
#turn off the datafeed
    DataSourceType='NULL';

EndIf;

If the MDX view returns no records I always get a fatal error "Error: Prolog procedure line (0): Unable to open data source "Plan"".
If I set booTempOn=0, then I do not see a view in PAW (after refreshing)
Anyway something seems off here.
If I have a traditional view that is empty the TI doesn't complain.
If I have a MDX view that is empty then I get a fatal error, additionally I have no way of testing to see if the view is empty to trap the fatal error.
Why do I not see the empty view in the object explorer when the ViewExists function is returning true?

I've been fiddling around with creating a union of my real query with a tuple that I know will return a cell but this adds a lot of downstream complexity in getting that record ignored when I do get results.

Anyone found a nice solution to this?



------------------------------
Steve Rowe
Technical Director, InfoCat
www.infocat.co.uk
------------------------------

#PlanningAnalyticswithWatson

I found an answer on the tm1forum

https://www.tm1forum.com/viewtopic.php?f=3&t=14797&p=73290&hilit=viewmdxset#p73290

Would be nice not to have to employ these workarounds.

The solution is 
1.  In the MDX do your cross join excluding the non-empty
2.  Then zero suppress the view using the command for the old style views.

Any views as to a performance it doing it this way around rather than doing a non empty cross join?

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

Original Message:
Sent: Wed April 29, 2020 05:10 AM
From: Steven Rowe
Subject: ViewCreateByMDX - Using as a datasource

Hi,
If I create a view with MDX and then assign it as a datasource to a view, this works well but...

If the MDX query returns no results then (as far as I can tell) I run into errors that I can not trap or suppress.

ViewCreateByMDX(sCubeName , sObjectName, sMDX, booTempOn);

#Check that the view got created (failed attempt to trap error)
If(ViewExists(sCUbeName , sObjectName)=1);

#Make the view the source
     DatasourceCubeview=sObjectName;

Else;
#turn off the datafeed
    DataSourceType='NULL';

EndIf;

If the MDX view returns no records I always get a fatal error "Error: Prolog procedure line (0): Unable to open data source "Plan"".
If I set booTempOn=0, then I do not see a view in PAW (after refreshing)
Anyway something seems off here.
If I have a traditional view that is empty the TI doesn't complain.
If I have a MDX view that is empty then I get a fatal error, additionally I have no way of testing to see if the view is empty to trap the fatal error.
Why do I not see the empty view in the object explorer when the ViewExists function is returning true?

I've been fiddling around with creating a union of my real query with a tuple that I know will return a cell but this adds a lot of downstream complexity in getting that record ignored when I do get results.

Anyone found a nice solution to this?



------------------------------
Steve Rowe
Technical Director, InfoCat
www.infocat.co.uk
------------------------------

#PlanningAnalyticswithWatson

There is an open APAR to have this fixed: https://www.ibm.com/support/pages/apar/PH16537

------------------------------
Tim Ryan
------------------------------

Original Message:
Sent: Thu April 30, 2020 03:47 AM
From: Steven Rowe
Subject: ViewCreateByMDX - Using as a datasource

I found an answer on the tm1forum

https://www.tm1forum.com/viewtopic.php?f=3&t=14797&p=73290&hilit=viewmdxset#p73290

Would be nice not to have to employ these workarounds.

The solution is 
1.  In the MDX do your cross join excluding the non-empty
2.  Then zero suppress the view using the command for the old style views.

Any views as to a performance it doing it this way around rather than doing a non empty cross join?

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

Original Message:
Sent: Wed April 29, 2020 05:10 AM
From: Steven Rowe
Subject: ViewCreateByMDX - Using as a datasource

Hi,
If I create a view with MDX and then assign it as a datasource to a view, this works well but...

If the MDX query returns no results then (as far as I can tell) I run into errors that I can not trap or suppress.

ViewCreateByMDX(sCubeName , sObjectName, sMDX, booTempOn);

#Check that the view got created (failed attempt to trap error)
If(ViewExists(sCUbeName , sObjectName)=1);

#Make the view the source
     DatasourceCubeview=sObjectName;

Else;
#turn off the datafeed
    DataSourceType='NULL';

EndIf;

If the MDX view returns no records I always get a fatal error "Error: Prolog procedure line (0): Unable to open data source "Plan"".
If I set booTempOn=0, then I do not see a view in PAW (after refreshing)
Anyway something seems off here.
If I have a traditional view that is empty the TI doesn't complain.
If I have a MDX view that is empty then I get a fatal error, additionally I have no way of testing to see if the view is empty to trap the fatal error.
Why do I not see the empty view in the object explorer when the ViewExists function is returning true?

I've been fiddling around with creating a union of my real query with a tuple that I know will return a cell but this adds a lot of downstream complexity in getting that record ignored when I do get results.

Anyone found a nice solution to this?



------------------------------
Steve Rowe
Technical Director, InfoCat
www.infocat.co.uk
------------------------------

#PlanningAnalyticswithWatson

Thanks for the update Tim, though I see it has been in the queue for a while...

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

Original Message:
Sent: Thu April 30, 2020 03:52 AM
From: Tim Ryan
Subject: ViewCreateByMDX - Using as a datasource


There is an open APAR to have this fixed: https://www.ibm.com/support/pages/apar/PH16537

------------------------------
Tim Ryan

Original Message:
Sent: Thu April 30, 2020 03:47 AM
From: Steven Rowe
Subject: ViewCreateByMDX - Using as a datasource

I found an answer on the tm1forum

https://www.tm1forum.com/viewtopic.php?f=3&t=14797&p=73290&hilit=viewmdxset#p73290

Would be nice not to have to employ these workarounds.

The solution is 
1.  In the MDX do your cross join excluding the non-empty
2.  Then zero suppress the view using the command for the old style views.

Any views as to a performance it doing it this way around rather than doing a non empty cross join?

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

Original Message:
Sent: Wed April 29, 2020 05:10 AM
From: Steven Rowe
Subject: ViewCreateByMDX - Using as a datasource

Hi,
If I create a view with MDX and then assign it as a datasource to a view, this works well but...

If the MDX query returns no results then (as far as I can tell) I run into errors that I can not trap or suppress.

ViewCreateByMDX(sCubeName , sObjectName, sMDX, booTempOn);

#Check that the view got created (failed attempt to trap error)
If(ViewExists(sCUbeName , sObjectName)=1);

#Make the view the source
     DatasourceCubeview=sObjectName;

Else;
#turn off the datafeed
    DataSourceType='NULL';

EndIf;

If the MDX view returns no records I always get a fatal error "Error: Prolog procedure line (0): Unable to open data source "Plan"".
If I set booTempOn=0, then I do not see a view in PAW (after refreshing)
Anyway something seems off here.
If I have a traditional view that is empty the TI doesn't complain.
If I have a MDX view that is empty then I get a fatal error, additionally I have no way of testing to see if the view is empty to trap the fatal error.
Why do I not see the empty view in the object explorer when the ViewExists function is returning true?

I've been fiddling around with creating a union of my real query with a tuple that I know will return a cell but this adds a lot of downstream complexity in getting that record ignored when I do get results.

Anyone found a nice solution to this?



------------------------------
Steve Rowe
Technical Director, InfoCat
www.infocat.co.uk
------------------------------

#PlanningAnalyticswithWatson