Filtering a table based on Row Level Security

[AlexanderRbt]

Hi everybody, 

This may be a tricky one.

I would like to create a filtered table based on a table of all users (users_all). This filtering is based on a permission table that is itself filtered by Row Level Security.

The permissions tables is called `groups`, the filtered table is called `filtered_users` and the original user table is called `users_all`.

The RLS correctly filters my permission table.

I use then this syntax to filter users_all based on the values in groups[group_id]

filtered_users = CALCULATETABLE(users_all,
                                KEEPFILTERS( users_all[group_id] IN DISTINCT(groups[group_id]) )
                                )  

However, although I can see that the groups[group_id] is correctly filtered in my table view, the resulting filtered table is untouched.

What am I doing wrong? Is it possible that the filtered table is generated BEFORE the RLS is applied? 

[lbendlin]

Usually RLS is applied to an outer dimension in your data model and then propagates through all the way to the fact tables.  No need for extra measures etc.

[AlexanderRbt]

I agree and this is how I usually use it. I have two permissions tables that are filtered by RLS. My users_all table needs to be filtered by both of these tables to create a superset (a person has to be in one OR the other table). I cannot seem to do that with the usual relationships in the data model view. If I create the two relationships, I only retain people who exist in both tables.

I tried to calculate a combined table with my RLS tables as filters but the filters are ignored as I described.