My customer has an old QS app (not my work;) with, among other tables, these 3 tables.
Two big facttables (Offers and Orders) with a lot of common fields
One dimension table with customers
While building the datamodel, after the script reload, QS:
Create a big syn-table
Delete specific rows of the dimension table. The rows that are deleted are the rows with unused key-values in both facttables. In our case the rows of customeres who does not have any offers or orders, but we still want to see those customers.
Is this a bug? Is there a way, beside get rid of the syn table, to fix this?
I'm further quite sure that there is no reducing of the data-set during the creation of the data-model. All records which were be loaded at this state will be there because there is no way on which they could be vanish.
Therefore please read my answers from above again - carefully.
In the end you may decide to design the data-model differently but just to check the that all loaded records are there concatenate all fact-tables into a single table and adding recno + rowno + an extra manual source-field. The dimension-tables are not necessary at this point. And now create a table-box within the UI and pull recno + rowno + Source and a few other relevant fields into it - and this check against your raw-data.
I just experienced this same issue and it had not been happening before. Someone suggested a data type issue or data changes on the back-end, but that doesn't really add up for me. Did you ever figure it out @qlikhalmar ?
Just go carefully through all my above mentioned hints. Without explicit statements no data will be removed and with recno/rowno it's simple to track all data.
