Data type problem since update to February 2019

[swiss49_ag]

Dear all

Since the update to Qlik Sense February 2019 most of our charts display the error "The chart is not displayed because it contains only undefined values.".

Our load script to gather data from SQL Server uses dynamic SQL and is thus rather complex. We think that Qlik Sense may be unable to properly identify the data type. Tags $ascii, $text, $numeric are all assigned to the columns, however the load script completes without any errors.

Has anything changed about the way numeric values are handled in Qlik in the February 2019 update? Something like strict handling of data types?

Best regards,
Matthias

[poojaBI]

Facing the same issue after Feb 2019 update. 

For temporary workaround , i have converted data type as  text / num whatever applicable in qliksense before joining.

We have rasied case with Qlik on this issue. Lets see what they say

[swiss49_ag]

Hi PoojaBi

Thanks for your answer - good to know this is not only on our side!
How do you define datatypes in order have QlikSense recognize it as Numeric for example?

Best Matthias

[poojaBI]

Sorry for the delay in response. In my scenario , i just have to wrap text function to keep them in synch 

So something like 

text( table1.column A) = text (table2.columnB) 

Both column have datatypes as varchar in database. But somehow qliksense Feb Release its treating them numeric in one table and text in another and not joining properly. So i have to apply this workaround.

[garystrader]

Did you receive anything from Qlik Support indicating if this is a bug, new works as designed feature, etc?

[Vegar]

One of the big issues reported in this community with the Feb 19 release is how Qlik Sense are handling numbers with leading zeros. Leading zeros are some times handled as digits, a ID=0080 will be read as 80. When performing an join on ID with 0080 in one of the table and ID 80 in the other Qlik Sense will find no match.

The workaround is to either put NUM(ID) as ID in both tables or TEXT(ID) in both tables to ensure identical values. In systems where the leading zeros are important you should choose the later.

[swiss49_ag]

Hi Garystrader

I'm trying once more to get this issue going...

As per info from our local Qlik representative, this is a change in behaviour, that is intended. So we're not really sure, which way to go now...
Best Matthias

[swiss49_ag]

Hi Vegar
Thanks for your post in June. I'm trying to get smart about the whole issue once more.

On our end, the workaround cannot be applied (even though it seems to be smart and straight forward), because we're loading this with a dynamic SQL loadscript. So we don't know the datatype then... So I guess we need to wrap our brain around it once more...

Problem not yet solved.
Thanks anyways!

Matthias

[marcus_sommer]

Maybe you could use the following (not only as workaround else as a general approach to format the data):

https://qlikviewcookbook.com/2016/07/touchless-formatting/

- Marcus

[silvacarlos]

Hi, 

We did an upgrade from Sept 2018 to June 2019 and we have the same problem, so I can understand until now this is not fixed or something. I've not checked Sept 2019.

Because of this problem we spent many time checking our scripts and doing the workaround that some people have already told you in this topic.

CS