I was recently trying to load 800+ records into my application for demo purposes.
I know eventually my client would need to load 20000+ records which has exceeded the qlik sense limit of 10000 cell data.
As such I was reading through the following articles to aid my situation.
While the data does get loaded initially, I does realised if I fool around with the current selection by selecting and deselecting values, eventually the backendApi would mix up and instead of retrieving the correct set of say for example 500 records, they retrieve 450 correct records and 50 duplicated records.
I know it's a little difficult to understand so I have attach the necessary files to replicate the issue for anyone who wish to try out.
In the sample, there is an qvf application that load the extension, the extension that will produce the behavior, a sample data file that is preloaded into the app and finally a screenshot of the duplicate I been mentioning.
To replicate, after loading the application in qlik sense, preferably launch it via browser such as chrome using
http://localhost:4848/hub and activate the developer tool.
Once there, there are 2 panels, extension is on the left and a table is on the right.
Simply select any value in VALUE2 column to start populating qlik sense current selection bar.
Then from the selection bar, go ahead and select all the values for VALUE2.
Now try to reduce them one by one.
Then add them back one by one.
You should be able to replicate the problem when the developer console is showing "duplicate - xxxxxxxxxx".
These "duplicate - xxxxxxxxxx" are the mentioned records that is duplicated and are replacing my actual selection data.
As such, layout.qHyperCube.qDataPages.qMatrix is not returning me my complete set of data.
Would totally appreciate some help on this matter, thanks!
P.S. The solution works perfectly when I set the read size to be 1000 rather than 20.
selection_bug_example.zip 341.0 K