Fix issues with 1.7.1 kedro-datasets release

[ravi-kumar-pilla]

With Kedro Datasets 1.7.1 release , we need to update the dataset imports and fix pytests

We will be pinning down the dependency of kedro datasets to 1.7.0 as a quick fix

[astrojuanlu]

Hi @ravi-kumar-pilla , what broke? Do you have a URL of a failing build?

[ravi-kumar-pilla]

Hi @ravi-kumar-pilla , what broke? Do you have a URL of a failing build?

Hi Juan, The circle ci builds are failing - https://app.circleci.com/pipelines/github/kedro-org/kedro-viz/10632/workflows/a8d8d61d-132a-4a48-8990-1ffa51bd422b/jobs/60837

[astrojuanlu]

I see - if I understand correctly, technically yeah this is a breaking change from 1.7.0 to 1.7.1 right? so maybe there's something to be done on the datasets side. cc @ankatiyar @merelcht

[ankatiyar]

Changing DataSet -> Dataset in the tests should fix this I think

[merelcht]

Yes that would fix it. I'm surprised that the tests started failing, because the aliases are still in place.. It might be because of hardcoded DataSet in strings though..

[ankatiyar]

Yeah, you can still create datasets with CSVDataSet(filepath) but the type of the dataset created would be CSVDataset and it's the assertions against hardcoded strings in the tests that are failing ->

assert transcoded_data_node_metadata.transcoded_types == [
            "pandas.csv_dataset.CSVDataSet"
        ]

[ravi-kumar-pilla]

Thank you for the suggestions. As a quick fix, we decided to pin down kedro-datasets version to 1.7 in the requirements file. We will be adding support for kedro datasets 1.7.1 and above in the upcoming sprints. Thank you