Issue about CALCULATETABLE

[CNENFRNL]

Dear gurus,

I came across an issue which I can't wrap my head around; so I turn to you for help. Thnx in advance!

The senario is fairly simple, which involves only one simple table of dates. I use the date column as slicer; then I'd calculate all dates later than the selected date.

I authored a measures with CALCULATETABLE, which produces a wrong answer.

Wrong = 
VAR __date = MAX ( 'Calendar'[Date] )
RETURN
    CONCATENATEX (
        CALCULATETABLE ( ALL ( 'Calendar'[Date] ), 'Calendar'[Date] >= __date ),
        'Calendar'[Date],
        UNICHAR ( 10 )
    )

Then I replace CALCULATETABLE with FILTER, which produces an answer as expected,

Correct = 
VAR __date = MAX ( 'Calendar'[Date] )
RETURN
    CONCATENATEX (
        FILTER ( ALL ( 'Calendar'[Date] ), 'Calendar'[Date] >= __date ),
        'Calendar'[Date],
        UNICHAR ( 10 )
    )

I understand that in general, context transition occurs while leveraging CALCULATETABLE. Here I didn't manage to figure out which context transit in the calculation. There must be some other reasons resulting in the unexpected answer.🤔

A mockup file is attached for more clarification. Thanks!

[Fowmy · Accepted solution]

@CNENFRNL 

Yes, CALCULATETABLE creates a context transition but FILTER doesn't.  In your case, if you look at the arguments in CALCULATETABLE, after you have applied 'Calendar'[Date] >= __date, you remove filters using ALL ( 'Calendar'[Date] ), that is the reason all dates are in the result. But if you apply VALUES to respect the filters then you can get the correct results with the same formula.

Corrected = 
VAR __date = MAX ( 'Calendar'[Date] )
RETURN
    CONCATENATEX (
        CALCULATETABLE ( VALUES('Calendar'[Date] ), 'Calendar'[Date] > __date ),
        'Calendar'[Date],
        UNICHAR ( 10 )
    )

________________________

If my answer was helpful, please consider Accept it as the solution to help the other members find it

Click on the Thumbs-Up icon if you like this reply 🙂

YouTube  LinkedIn

[CNENFRNL]

Thank you very much, @Fowmy for such a detaled explanation!