Tate Bowman

@marcorus OBT()

It's Saturday. Name a DAX function that doesn't exist but absolutely should.  Wrong answers only. 😎 I'll start: MAKENOMISTAKES().

@kurtbuhler To think that a 3rd-party tool has achieved such a first-class development experience for a platform is really remarkable. We have come a long way from the SSDT days.

The Tabular Editor CLI is now available in limited preview! Use it from the terminal, with a coding agent, or for automation, scripting, and CI/CD. I've been having a blast testing this tool - it gives you or an agent headless access to the Power BI pro developer features of TE3. Link: tabulareditor.com/blog/introduci…

@ehansalytics @kurtbuhler @data_ben I think this is generally done to appease a model author’s OCD-tendencies rather than a “best-practice”. PBI needs a “Table Ordinal” property similar to the Column Ordinal property that exists today (and client tools conveniently ignore). Though that may not help those with OCD😅

@kurtbuhler @data_ben I cannot convince some people to stop using dimDimension and factFact patterns. Nerd speak is not good in a semantic model.

How you name things - workspaces, items, tables, fields, folders, and even variables - is very important. Not just for organization / consistency, but for better results using AI for dev or conversational BI. Naming conventions for a semantic model: tabulareditor.com/blog/naming-co…

@kurtbuhler I really like this. Now if only we could “frictionlessly” rename measures without breaking reports (which works for composite models but not for live-connected reports!), cleanup activities would be a no-brainer.

@JohannesVink docs.tabulareditor.com/en/how-tos/pow…

And after years there is still a scary long list of limitations. #limitations" target="_blank" rel="nofollow noopener">learn.microsoft.com/en-us/power-bi…

And this people is why I am very very careful with all kinds of XMLA write features in Power BI... It all works, until suddenly...

“DaX dOeSn’T sUpPoRt ReCuRsIoN” You don’t need it. TimeSeries.MovingAverage (now on DAX Lib) contains several different flavors of MA’s, most interesting of which is Exponential. Traditional EMA calcs require recursion, but we just rethink the possible! daxlib.org/package/TimeSe…

@marcorus It seems SUMMARIZECOLUMNS does not respect USERELATIONSHIP inside a surrounding CALCULATETABLE, whereas the SUMMARIZE equivalent does. Am I missing something, or is this one of the exceptions?

Yes, you can use SUMMARIZECOLUMNS in a DAX measure, BUT... But only if you follow the best practices in this article + video. Spoiler: Goodbye ADDCOLUMNS / SUMMARIZE - well, with some exceptions... #dax #powerbi sqlbi.com/articles/summa…

@kurtbuhler That’s it. We need this turned into a Magic Puzzle Company puzzle now.

There's also a dark mode version 😉

@GuyInACube @MicahDail @marcorus Wait, are we getting Custom Calendars this month?!

@MicahDail @marcorus You will like it Micah... :)

New DAX functions just dropped y'all! @marcorus, what exactly is the '<Calendar>' parameter on DATESWTD()?? dax.guide/changelog/

@ehansalytics @_christianWade He’s working on it! 🎉 #issuecomment-2964714712" target="_blank" rel="nofollow noopener">github.com/microsoft/Anal…

@_christianWade Now we just need the third party tools to get updated to support this. ALM Toolkit isn't yet aware it is allowed to make these changes.

Now it is official. Desktop is officially hardened. What does that mean? The real question is: what does that enable (in addition to the obvious things like unlimited external tool capabilities, MCP server compatibility, AAS migration with PBI supported tooling, full semantic model metadata compatibility, automated creation of ...)? ... maybe I should let the community answer that. Did anyone see this coming? Thank you @RuiRomano 🙏

@_christianWade Long live BISM Normalizer!

@marcorus Can UDFs reference other UDFs?

I usually do not write about announcements and new features until we have had time to try and test them in the real world. However, there are always exceptions, and some of the announcements at the Microsoft Fabric Conference 2025 fall into this category because I have worked with them enough to provide hands-on feedback. In short, these are the topics I am covering in my latest post: - Direct Lake and Import mode - Calendars in DAX - User Defined Functions (UDF) in DAX Read more: sqlbi.com/blog/marco/202… #dax #powerbi #fabric #datalake

@marcorus Promise no April Fools?

We have 600 people already registered for the DAX session at #fabcon, during which Jeroen ter Heerdt and I will announce and demonstrate new, exciting features for DAX, which are coming soon. Yesterday, my session was fully packed because the room was too small, but today, the room is very large. Everyone will have a seat! 1:45pm, Premiere Ballroom 313, 3rd floor. If you are attending the Fabric Conference, you do not want to miss this session! See you there!

YYYYMMDD - File Description.xlsx

@marcorus @kurtbuhler I am sick of Figma for this reason.

Big claims require big proof. Not Figma demos. (thanks @kurtbuhler)

Study The Brick with me🤓 I got carried away building diagrams with ReactFlow and built a DAX query plan visualizer. - Works with both JSON and plain text plans - Share plans using a shortcode - All processing/parsing happens in the browser

@cwebb_bi @marcorus @MicahDail @notaboutthecell @PhilSeamark Which leads us to creating our own “aggs” with specific grouping sets, really complex DAX to set correct filter grain, and we’re left with a really rigid solution. If you could re-aggregate distinct counts based on arbitrary selections, I think big data customers would go bonkers

@cwebb_bi @marcorus @MicahDail @notaboutthecell @PhilSeamark Our users absolutely would be comfortable with approximates, and you could make the precision configurable on creation of the HLL sketch. The reality is, while approximate distinct count is faster than the exact, it’s just not fast enough to blow users away like other aggs.

@marcorus / @cwebb_bi / @notaboutthecell / @PhilSeamark have you ever messed with HLL++ sketches? Distinct count operations at our scale are very challenging. A couple challenges to get this working: 1. While Vertipaq does have a binary data type, Power Query no likey, and there's all kinds of hacks to split binary into 30k text upon load instead of retaining the binary type. We would need to retain the binary data type. 2. It would be absolutely dope to actually have this in memory. I understand it doesn't compress and could blow the memory of the model. 3. Even if we chose not to import, this function could be a DQ function (think APPROXIMATEDDISTICNTCOUNT ) operating on the binary column type held in the upstream system. 4. WE NEED HLL SUPPORT IN DAX. Kusto has this support, but not DAX. It would be absolutely bonkers cool if we could invoke a function from DAX to merge sketches from the upstream system and just spit out the approximate distinct counts. We could then unleash our data engineering teams to product HLL sketch aggregates and start knocking out ALL of our distinct count problems. Please help. DAX is missing a good solution on distinct counts at scale. Also, I don't see why this couldn't be a defining feature of and only supported in Direct Lake , to add more glam to Fabric adoption.

@MicahDail @marcorus @cwebb_bi @notaboutthecell @PhilSeamark My #1 #PowerBI request right here