![]() ![]() Testings) - the usual format strings of normal measures are properly shown If *all* FormatStringExpression of *all* Calculation Items areĮmpty, then the behaviour of measures is unchanged (at least in my When Calc groups are added to the model, yes, the measures become I have the same issue here, and I have discovered something that might I hope this helps shed some light, and yes, Microsoft should solve this! Therefore, in my case so far I am able to add Calculation Groups and Items without and issue to an existing model (phew!), but not use the custom format feature. So in my case, it looks that the problem does not come from just 1500, or from adding calculation groups, but from adding a single FormatStringExpression to the model. If afterwards, I add just one FormatStringExpression to ANY of the Calculation Items, (including SELECTEDMEASUREFORMATSTRING() or any other), then the bug arises and graphs in PowerBI do not respect their source formatting (matrix is OK as you have reported).If all FormatStringExpression of all Calculation Items are empty, then the behaviour of measures is unchanged (at least in my testings) - the usual format strings of normal measures are properly shown in graphs, business as usual, all happy. ![]() When Calc groups are added to the model, yes, the measures become variant.The thing is, in my case, I find the following: I have the same issue here, and I have discovered something that might help in the struggle. Alternatively, they should query the model using XMLA and consume the returned cellset, but this probably requires much more effort. That’s because Excel also uses XMLA, but since it uses XMLA with MDX it’s also able to get the format string per cell.Īs to the fix: Microsoft either needs to make sure that all visuals query the hidden format string measures for every measure that is being used in the visual, as well as ensure that the visual actually applies that format string when rendering the value. You probably don’t see this issue with the format strings in Excel. I’m guessing Power BI uses the latter or another similar protocol that returns a flat dataset. On the other hand, when querying the model with an OLEDB connector, we get a flat dataset as the result, where its not possible to have differing data types per cell. Invoke-AsCmd actually uses XMLA to send the command to the model, meaning the returned result is a cellset, which explains why the data type is specified for each cell. Maybe knows the internals in more details. If you have any calc items that doesn’t follow this pattern, they have to decay all measure data types to variant.Īs to the issue with format strings, here is some speculation on what I think is going on. These patterns are unfortunately not documented, as far as I know. ) where the data type is guaranteed to not change. It happens when you have calculation items that do not follow certain patterns, such as CALCULATE(SELECTEDMEASURE(). The issue with measures changing their data types to variant is familiar. It is a shame they didn't implement Format Strings in the same way they did for Data Types as this is provided as metadata with the resultset - although now you have to check per cell of results, again instead of being able to just get the datatype from the column definition. So now the query tools need to write queries that return the format string for this specific context, rather than just being able to take the format string from the measure. Suddenly every measure now has 2 measures with the same name (the measure itself and also the measure containing the format string) - you see it in both power bi and dax studio In the basic example with a Date Calculations table with a "YOY %" then yes, you have to provide a format string to have this present the number as a % instead of the selectedmeasure's formatstring.Īs soon as you add a calculation group to a model, Analysis Services dramatically changes the model in how it provides the queries with the format string to apply. It makes no difference, Power BI just isn't looking in the right place for these non-table visuals Yeah we tried all of that when we initially saw the problem. ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |