Fix for the Azure DataLake connector

[CasperLehmann]

When I use the ADLS connector from Power BI Desktop, everything runs smoothly. But as soon as I upload the model, updating stops working and in stead I get the error:

We cannot convert the value null to type Logical. Table: {TableName}.

I found that the problem only appears with the default parameters of the connector.

When you use the Azure Data Lake connector Power BI creates your query for you. The resulting query, running through DataLake.Contents is then supplied with default parameters that (at the time of writing) break the evaluation once uploaded to the Power BI service. The problem is the second param, the empty record []. Exchange it for a null, or simply delete it.

Bad:

DataLake.Contents("adl://adlstest01.azuredatalakestore.net/data/examples/entries.csv", [])

Good:

DataLake.Contents("adl://adlstest01.azuredatalakestore.net/data/examples/entries.csv", null)

Perhaps the default perameters used to initialize the connector could be could be set to null, but having the integration api fixed would be nice too.

[v-yuezhe-msft]

@CasperLehmann,

This is a known issue, and it will be fixed in Power BI Desktop August release.

Regards,
Lydia

[CasperLehmann]

Seems this is still an issue. The default parameter is a dictionary [PageSize=null], which chrashes the Power BI Service (even if it works in Desktop). Removing the parameter seems to solve the problem.

I havn't had a very consistent experience with this, but sometimes setting it to a simple null has also worked. Mind you, this problem is specifically on the service.