DAX Operator

Hi Anonymous ,

 

They are both the same but one A will be more cost efficient.

 

See this video by Jeffery Wang - fro the team who created the DAX Engine

https://www.youtube.com/watch?v=bJtRB86n9tk

 

The difference is explained in 37:540 in the video.

 

Regards,
Harsh Nathani

Appreciate with a Kudos!! (Click the Thumbs Up Button)
Did I answer your question? Mark my post as a solution!

Anonymous Both measures are different, and the first one is incorrect here is why:

 

When you write a predicate/boolean logic in the filter argument of CALCULATE then internally it expands to a FILTER and ALL construct. Product[Color] = "Red" becomes 

= FILTER ( ALL ( Product[Color] ), Product[Color] = "Red" )

In the first option the two predicates refer to 2 different column of Products in the same filter argument of CALCULATE, i.e. Color and Size, since both of them will be expanded by the engine, but engine won't be able to create a combination of unique existing Color and Size from the predicates, that's why you are not allowed to use predicates that refer to more than 1 column in single filter argument of CALCULATE, Product[Color] = "Red" && Product[Color] = "Blue" is fine and won't throw an error

 

if you want to use 2 columns in the same CALCULATE filter argument you will have to use this:

=
CALCULATE (
    [Measure],
    FILTER (
        ALL ( Product[Color], Product[Size] ),
        Product[Color] = "Red" || Product[Size] = 50
    )
)

 

in the second option the naked reference to the Product table will respect any filter coming from anywhere in the report, but when you use predicates, ALL removes those filters. so in case if you wanted to have a 2 column filter while ensuring the new filter is intersected with the filter outside CALCULATE without overwriting it, then you need to use KEEPFILTERS.

 

=
CALCULATE (
    [Measure],
    KEEPFILTERS (
        FILTER (
            ALL ( Product[Color], Product[Size] ),
            Product[Color] = "Red" || Product[Size] = 50
        )
    )
)

 

Also don't use a table filter where column filter will do the job it is much more efficient as there is no need to apply every column of the products table to the filter context when it is not needed, also if you are unaware of expanded tables then the values returned by a measure could be surprising.