Slicer and table only filters on random items -- Bug?

[adahlin]

I'm creating a report that has a set of user selectable transactions in a table, each of which have an asset associated with them. When you click on a transaction in the table, a matrix and line chart display the associated asset's performance over time. Each transaction is associated with a Trading Group (TG) and I have a slicer to filter on this. Pretty straightforward. Basically a Transaction has an Asset has a Performance History.

The problem is that depending on which columns I display in the table, it will randomly filter some items but not others. I've tried changing relationships and other things, but it still has the same bizarre behavior. See screenshots below for a walkthrough of the madness.

After a LOT of troubleshooting, I've narrowed it down to this. If I have a datetime field AND the same field the slicer is on (TG), most of the items won't show their history if they're selected, but a few will. If an item isn't selected, the items that don't work when selected have their history shown. It's like it's failin on drilling down.. most of the time. I can't find any difference in data between rows that work and ones that don't: they appear identical in form.

If I change the datetime field to a Date Hierarchy instead of a date, it works. If I remove the datetime field from the table, it works. If I remove the TG field from the table, it works.

All the tables and charts have their default settings; nothing fancy. The only thing I changed is the TT_Order <-> OrderPricingHistoryContext relationship filters both ways.

Thoughts?

 

[v-jiascu-msft]

Hi @adahlin,

This could be the magic of Context. Please see the demo below. If this isn't the cause, please share your pbix file or a dummy sample.

Best Regards.

Dale

[adahlin]

Hi @v-jiascu-msft,

I don't think that's the problem, but that's something really good to be aware of; I'm sure I'll do exactly that at some point in the future!

I can't share the pbix directly due to privacy concerns, but I'll see if I can create replicate the problem with some sample data.

Thanks for the reply!