Calculated Column versus Measure syntax

[dicken]

Hi,
Can someone explain why in a cumulative total by date in a calculated column the filter needs the whole table, as opposed to a measure,
where the advice seems to be use the least number of columns so in this case one ; date. But if I just put filer Table[ Date] in a CC I
just get back the row value? what's happening?
Calculated column Filter entire table

 CALCULATE(SUM(Table1[Sales]), FILTER(ALL(Table1),
                                                                           Table1[Date]<= EARLIER(Table1[Date]) ))

Measure just date ;

RTC :=
VAR sdate =
    MAX ( Table1[Date] )
RETURN
    CALCULATE (
        SUM ( Table1[Sales] ),
        FILTER ( ALL ( Table1[Date] ), Table1[Date] <= sdate )
    )

 

[Matt Allington]

When you use calculate in a calculated column, the row context (always exists in a calc column) is converted to a filter context. Every column in the table is converted to a filter context, so you need to remove the filter from all columns, aka, the table.
To demonstrate, try adding all (or simply more) columns to a visual (rows) and then see what your measure does.
Also try a calculated column without a calculate.
test column=SUM(Table1[Sales])

 

[dicken]

When you use calculate in a calculated column, the row context (always exists in a calc column) is converted to a filter context. Every column in the table is converted to a filter context, so you need to remove the filter from all columns, aka, the table.
To demonstrate, try adding all (or simply more) columns to a visual (rows) and then see what your measure does.
Also try a calculated column without a calculate.
test column=SUM(Table1[Sales])

Thanks, that does make sense, I think, adding more columns to the source table didn't make a difference to the measure,
I didn't expect it would. Should it have ?

Richard.

 

[Matt Allington]

Adding new columns to a table should normally not make a difference to a measure unless it is somehow referenced in the measure. That doesn't mean it is good practice to add more columns - in general it is considered bad practice (although there are always exceptions) https://exceleratorbi.com.au/calculated-columns-vs-measures-dax/