Changes to Dataset cause error in Live Connect

[yamayancha]

Hello everyone,

we have set up one main report with the underlying dataset in Power BI service for development purposes and multiple uses create their own reports which connect to this dataset. After some discussion we have changed the original model to fulfill style guidelines regarding table names and measure names and have updated the original dataset.

Now the additional report that has a live connection (or direct query) to the original dataset reflects those changes within the model - but all visualizations are no longer functional as they are still looking for tables and measures with the original names.

If changes in the original dataset cause the connected reports to stop working kind of defeat the purpose of such a live connection or direct query to an existing dataset. The goal was to be able to work on multiple reports at once without having to repeat changes to the original dataset in every single one of them as this dataset is supposed to be continuously developed. Is this something that is working according to plan or could this be changed?

[collinq]

Hi @yamayancha ,

I would argue that this is working as planned.  And, it is working for you in that when you changed the original dataset it flowed through to all of the reports and it was all updated without intervention by each report writer.

Where it does "break" the reports is in the visualizations that are looking for specifically named columns from specific tables.  Any DAX measures or columns will also break as they are looking for specific columns.  The visuals are looking for the literal column/measure that it was using before the change.

So, if it does break that means that you either removed a field/column and/or renamed it.  The solution is easy - the report writer(s) just have to go to the visual and select the new field.  OR, they have to go to their DAX formulas (if any are local and not in the dataset) and modify them to the new field name.

[yamayancha]

Hi collinq,

thank you for your reply. I understand now that this is how it was intended but what I fail to understand is, why it was intended to work this way. Would it not be much more user-friendly if any visuals or measures would also auto-update when the connected dataset changes? If you have multiple reports connected to the same dataset and some measure is renamed in the original dataset, now you have to go through every single report to change the visualizations that use this measure.

To me, this kind of defeats the purpose of connecting to a dataset as I now have to notify every single report author who uses my dataset for their reports and they have to update their reports before any viewer opens them and only sees error messages. I had hoped this was a bug that was going to be fixed as this revelation will definitely not help to raise my colleague's enthusiasm when working with Power BI.

[collinq · Accepted solution]

Hey @yamayancha ,

I can't speak for the Microsoft developer mindset but I see both sides of this issue.  For example, if I have "renamed" a field in the visual (and just in the visual) I may want that to be the display name no matter what the real field name is.  But, if the real field name is changed, I might actually want to use the new field name and not my version of it and this gives me a chance to double check that.  Or, I may have my own terminology and don't care what the field 

Of course, the other side of the coin is as you state - if I change it in the  dataset I want to flow through to everybody that way.  

I guess part of it comes down to the business philosophy - is the dataset the "official" terminology?

All that said - it might be good for you to create an idea (or find one) for this issue.