I am trying to fetch the rows from which I don't need duplicate rows.
in SQL I can use distinct but if I use distinct in OData SQL Server query filter it is not allowing...
what is the reason ?
Hi friend what I know is that in ODATA you can do this operations
System Query Option $filter
System Query Option $orderby
System Query Option $top and $skip
System Query Option $count
System Query Option $expand
System Query Option $select
System Query Option $search
Proud to be a Flownaut!
Hi @ShivaPatpi ,
Could you please show us how you configure specific actions?
Best Regards,
I used this work around
but my question still remains the same:
Why should I get duplicate values from SQL
later on why should I do additional work to remove the duplicates by setting the variable and followed by union.
why can't we have a feature request to get rows from SQL in that ODATA select statement itself ?
Check out new user group experience and if you are a leader please create your group
See the latest Power Automate innovations, updates, and demos from the Microsoft Business Applications Launch Event.
User | Count |
---|---|
45 | |
42 | |
41 | |
40 | |
32 |
User | Count |
---|---|
86 | |
85 | |
58 | |
50 | |
41 |