Issue with Dataflow Gen2 with date variable

Hi RamonNooijen ,

The error message is actually giving the clue:

We cannot apply operator < to types Table and Date

That means WM is not a single date value at runtime – it’s still a table (or query result) when the dataflow engine tries to fold the query and push it to the Lakehouse / DW.

When you hard-code WM = #date(2025, 12, 3) it works, because WM is clearly a scalar date.
When you load it from SQL, your WM step is returning a table (for example one row, one column), not the cell value itself.

You need one extra step to turn that table into a single date before using it in Table.SelectRows.

A common pattern looks like this in Power Query (M):

// Query that gets the watermark table from SQL
WMTable = Sql.Database("server", "db"){[Name="WatermarkTable"]}[Data],

// Take the first row and the Watermark column as a scalar value
WM = Date.From( WMTable{0}[WatermarkDate] ),

Then in your filter step:

FilteredRows =
Table.SelectRows(
IssuesTyped,
each [updated_at] >= WM
)

Key idea:

• WMTable = table returned from SQL

• WM = single Date value extracted from that table

Right now you’re comparing [updated_at] (a date) to WM (a table), which is why you get the “Table and Date” error only when it actually runs in the service.

Once WM is a true date value, your dynamic filter will work the same way as the hard-coded example.

– Gopi Krishna