Daniil Daniil on 13 Dec 2018 11:21:02
Dataflows should have parity with Power Query when it comes to data types.
Currently, dates are converted to datetime and fixed decimal numbers are converted to decimal numbers. Time does not work at all.
I haven't tested all other data types, so I don't know if these three types (date, time, fixed decimal number) are all we need.
- Comments (9)
RE: Dataflows data types parity with Power Query
Thanks vor this very important feature request!
btw. I wrote a blog article how you can still use durations in Power BI dataflows (but not export them)
https://www.brunner.bi/post/how-to-fix-duration-data-types-in-power-bi-dataflows
RE: Dataflows data types parity with Power Query
Trying to connect MS Project Online to CDS and the datatype duration is converted into a number, with no known recourse to my knowledge. This is a definite impediment across the entire power platform.
RE: Dataflows data types parity with Power Query
Additionally there is a problem with the display of Whole Numbers. They have a data type of Int64.Type, and they are stored internally correctly, but they are displayed rounded as if they are doubles (i.e. to 15 significant figures), as can be seen by the following which uses the maximum permissible value of an Int64:
let
Source = Table.FromRecords({[LargeInteger = 9223372036854775807]}, {"LargeInteger"}),
ToInteger = Table.TransformColumnTypes(Source,{{"LargeInteger", Int64.Type}}),
ToText = Table.TransformColumnTypes(ToInteger,{{"LargeInteger", type text}})
in
ToText
The "ToInteger" and "ToText" steps show different values when used in Power BI dataflows.
RE: Dataflows data types parity with Power Query
I was trying to develop a dataflow that returns a table with columns containing records or lists. You need to convert these complex types to text by encoding them. This is way too cumbersome.
RE: Dataflows data types parity with Power Query
Much needed, especially date is needed. After referencing a entity in another dataflow or in pbi desktop it converts back to datetime..
RE: Dataflows data types parity with Power Query
'@Anonymous the data type parity issue is not fixed, as far as I'm aware. So you get DateTime instead of Date at the moment.
What I referred to on the Issues website was the locale issue: there were errors when using US/Rest of the world date formats (MDY vs DMY). Here is the link for anyone curious: https://community.powerbi.com/t5/Issues/Dataflows-sets-data-types-incorrectly/idi-p/575824
RE: Dataflows data types parity with Power Query
'@Danill, it seems I'm still experiencing the wrong Date type issue - a field appears to be correct Date type field in Data Flow, however, when loaded the data flow into PBI Desktop, such filed becomes DateTime field (12:00:00.00 is automatically added).
I read through your previous post regarding to this issue and the status was "fixed".
RE: Dataflows data types parity with Power Query
Time has since been partially fixed -- now it is converted to DateTime.