Tag Archives: Microsoft Fabric

Less GUIDs: Relative Current Workspace & Items By Name

• , , ,

Power Query dataflows like to use GUIDs to reference workspaces, lakehouses and warehouses. With a GUID reference,  you are authoritatively referencing a specific resource, which is good…but working with GUIDs can entail complications, both for humans and CI/CD, which is not so good.

I know what lakehouse “SomeLakehouse” is, but I’ll have to look up what lakehouse “fb7322d6-781e-4e0a-97b1-1eec69a5547b” is. On the source control side, I find that, due to how GUIDs tend to be intermixed with other code in Power Query, Git often incorrectly wants to merge dev workspace GUIDs to test, or test-specific GUIDs to prod. If not caught by careful human PR/merge review, such inadvertent merges leads to bad things happening.

Can we use names instead of GUIDs? That would resolve at least part of the challenge. Our code could refer to “SomeLakehouse” instead of a GUID (nice!).

However, switching from GUIDs to names still leaves the need to update workspace references when a dataflow is promoted across workspaces (e.g. from dev -> test -> prod.) Is there a way to reference the current workspace, regardless of its name?

The answer is yes to both! Using names instead of GUIDs isn’t new, but hasn’t often been done. On the other hand, relative workspace references is a brand-new, in preview feature.

Let’s explore both, in code!

Continue reading →

What’s With WriteToDataDestination?

• , , , ,

If you examine the refresh details for a Power Query gen2 dataflow, you’ll notice that the activity names which are displayed are identical to the query names in your dataflow except that each has a “_WriteToDataDestination” suffix appended to it. What’s going on with this?

Let’s see what we can uncover! Time for some sleuthing….

Continue reading →

Time Values from Dataflow to Power BI via DirectLake

• , ,

Seems simple enough: The table, produced by your Power BI gen2 dataflow, contains a time column. You want to output this table to OneLake, then have Power BI read from it using a DirectLake-powered semantic model.

Sounds easy, but not so fast! When you try to set up the dataflow’s output mappings, columns of type time are not supported for OneLake destinations. Ouch!

So, what do you do?

Try leveraging the fact that Power BI’s time values are, behind the scenes, actually datetime values. For proper “type time” behavior, Power BI expects that the date part of each datetime value is set to Power BI’s epoch date (the date it considers day 0, which is 1899-12-30) and it also needs to know that the column is expected to have time behaviors.

Continue reading →