Salesforce plugin - dataset creation

dataiku / dataiku-contrib

Public repository for DSS plugins

Apache License 2.0

101 stars 82 forks source link

Salesforce plugin - dataset creation #84

Closed BGermonpre closed 6 years ago

BGermonpre commented 6 years ago

I'm not able to create a dataset from the Salesforce plugin when I choose the option "Readable with columns". More precisely, I cannot build the schema.

Error message: Test failed: java.lang.IllegalStateException: Expected BEGIN_OBJECT but was NULL at line 1 column 16 path $.schema, caused by: IllegalStateException: Expected BEGIN_OBJECT but was NULL at line 1 column 16 path $.schema

I can see the data in preview and it also works when I select "JSON" as the format, so I assume my connection has been set up correctly.

Anyone knows how to troubleshoot this? @jereze?

Thanks!

jereze commented 6 years ago

Hi Benoit,

Indeed, it seems that the connection works. Your problem seems specific to the output format with all columns.

I'm not able to reproduce. Can you please share the following information?

Which version of Dataiku DSS?
Which version of the plugin?
Which dataset?
What data are you fetching from Salesforce? If you try another Salesforce object or another plugin dataset, do you still have the issue?
How many columns? Anything special in the columns names? A special character?

As a workaround, you can use the JSON output format and then use a prepare recipe to parse the JSON.

BGermonpre commented 6 years ago

Thanks for the quick reply.

DSS version: 5.0.0 (free edition)
Note: I'm running on Windows via Docker
Plugin version: 1.1.0
I have the same issue for all types of data that I'm trying to fetch (object list, records of specific objects)
Example: Opportunity dataset contains 45 columns and 934 rows. There is nothing special/unique about them.

jereze commented 6 years ago

Can you upgrade to 5.0.1 or 5.0.2? I know that a bug around plugin datasets was fixed. It might help.

BGermonpre commented 6 years ago

Thanks for the info! I will use your workaround solution for now. It seems very easy and I'm only testing some functionality for now, so it will be sufficient.