I'm not looking for a solution to this issue, just confirmation that it is happening for others so that it can be addressed by Qlik in a future release. I have already implemented a solution.
Basically we have customers with assets (IT equipment) that are either covered by a warranty or not. We're creating a table that is sorted with the customers with the most assets at the top and if two customers have the same number of assets it should sort by the most covered assets next. So, I created a table to do this but am noticing discrepancies with the sorting within the table.
Attached is a simplified version of my data. I loaded the data to two tables to test things out, simply just switching the order of the customers with the same asset count. I then duplicated a front-end table for each back-end table and I'd expect them to both look identical, but alas they are not - the YesFirst table is correctly sorted, the NoFirst table is not.
I furthered my testing by creating two additional front-end tables with a Total column that counts both covered and uncovered assets. When I add this column and choose to sort by it first instead of by the expression in the Customer field both sort correctly. So, it seems to me that sorting by expression sorts properly until it finds two rows with the same value/count and then it defaults to load order rather than moving onto the next sort criteria.
Again, I'm not looking for a solution to the issue, just hoping that other people are seeing this behavior so it can be addressed. For reference, we are using Qlik Sense 3.2 SR2.