Azure SQL Database table - "Shape of passed values is (1, 100), indices imply (15, 100)"

[Keshav Malani]

I'm testing out using Azure SQL Database offering and am able to connect and link to tables (detects columns accurately) but when I got to SQL Lab to even preview the table, I get this error: "

Shape of passed values is (1, 100), indices imply (15, 100)". My table has 15 columns but no data is coming for them

I get no data returned when I'm in the Slice creation view either, no matter what visualization and setting I choose.

I had the issue on 0.17.4 and have upgraded to 0.17.6, and still have it.

Note: Issue is only coming up for SQL Database - mysql, redshift, sqlite are working fine

[Prathap Reddy]

Even I was facing the same issue in 0.17.5 with sqlserver and I faced it even with sqllite (i.e default database that ships with superset).

[Maxime Beauchemin]

Please report issues on Github, it's a better place than the mailing list to keep track of progress. 

It would really help if you can provide a stack trace and/or describe a way to reproduce with one of the common backends (MySQL, Postgres, sqllite).

As for Azure we currently don't have an entry for it in the docs as and in the module `db_engine_specs.py` where we specify specific logic or configuration that Superset requires that tend to diverge from an engine to the other and to which SQLAlchemy doesn't provide a common interface for. It may be easy to add a new entry for Azure that derives the class of another similar engine. 

You'll understand that Airbnb can't really take on fine-tunning and maintaining code for databases we don't have access to. We're looking for committers that will do that for individual database engines. For instance someone offered to help out on ironing out Redshift recently. We'll have to rely on people in the community to make sure all database engines all well supported.

Max

[Prathap Reddy]

Thanks Max.  Its true and humongous task to support each and every database.  

I actually debugged the issue in 0.17.5 and issue is during creation of Pandas DataFrame somehow it is not intercepting list of tuples passed to it.  If I hardcode the same values it's working properly but the data instance is not intercepted properly.

cdf = dataframe.SupersetDataFrame(pd.DataFrame(list(data), columns=column_names)) 

Thanks,

Prathap