Superset: Create Datasets Without Auto Chart Creation

The Friction of Automatic Chart Creation in Superset

Superset, a powerful open-source data exploration and visualization platform, offers a streamlined experience for users to dive into their data. One of the core functionalities is the ability to create datasets, which serve as the foundation for all subsequent analysis and visualizations. However, a common point of friction has been observed in the current workflow: upon creating a new dataset, users are automatically redirected to the chart creation page. While this might seem efficient for users who always intend to visualize their data immediately, it presents a significant hurdle for others. Many users create datasets for reasons beyond immediate visualization – perhaps for organizational purposes, to set up batch data processing, or simply to have a dataset ready for future exploration. The current enforced redirection adds unnecessary steps, forcing these users to either cancel the chart creation or navigate back manually to the dataset list. This enhancement aims to introduce flexibility into the dataset creation process, empowering users to choose their next step after successfully creating a dataset.

Understanding the Current Behavior and Its Limitations

The current behavior in Superset dictates that once a user navigates to the 'Add Dataset' page, selects a database and a table, and clicks the 'Create' button, the system not only saves the dataset but also immediately initiates the chart creation flow. There is no inherent option provided to simply create the dataset and remain within the dataset management interface. To illustrate this, consider the reproduction steps: a user accesses the 'Add Dataset' page, chooses their desired database and table, and clicks 'Create'. The immediate consequence is a redirection to /chart/add/?dataset={dataset_name}, the dedicated page for crafting new charts. This means that if a user's intention was merely to add a dataset to their catalog for later use, they are now presented with a chart editor they didn't intend to use, requiring them to back out of the flow. This can be a frustrating experience, especially for less experienced users or those performing routine data management tasks where immediate visualization isn't the goal. The absence of a choice here reduces the platform's adaptability to diverse user needs and workflows, making it less intuitive for a portion of its user base.

The Expected Behavior: Empowering User Choice

To address the limitations of the current workflow, the expected behavior is to grant users more control over their actions post-dataset creation. Instead of a single, unyielding 'Create' button, we propose transforming it into a dropdown button. This button would offer two distinct and clearly labeled options:

1. Create and go to chart creation: This option would preserve the current default behavior. Upon clicking this, the dataset would be created, and the user would be seamlessly redirected to the chart creation page, catering to those who wish to visualize immediately.
2. Create dataset only: This new option would allow users to create the dataset and then be redirected back to the dataset list page. This provides a clean exit point for users who have completed their primary objective of adding the dataset to their inventory.

The interface should clearly present these choices, ensuring that the 'Create and go to chart creation' option remains the primary and default action to maintain backward compatibility and ensure that existing user workflows are not disrupted. This approach balances the need for introducing new functionality with the importance of a stable and familiar user experience for established practices. The acceptance criteria are meticulously defined to ensure the successful implementation of this feature: the create button must be converted to a dropdown, the primary action must retain its existing functionality, the dropdown must include the new 'create only' option, successful selection of 'create only' must redirect to the dataset list, existing tests must pass, new tests must cover both navigation paths, and the button's disabled state must be maintained when no table is selected or when the dataset already exists. This thoughtful design ensures a more intuitive and efficient user journey.

Step-by-Step Verification of the New Workflow

To ensure the successful implementation of this new feature, a clear verification process has been outlined, allowing users and developers to confirm that the desired functionality works as expected. The manual testing steps are designed to be straightforward and cover all key aspects of the new 'create dataset without chart creation' option.

Firstly, users should navigate to the Add Dataset page, just as they would in the current workflow. The next step involves selecting a database and a table, which triggers the crucial change in the user interface. At this point, users should observe that the 'Create' button is no longer a simple, singular action. Instead, it should now appear as a dropdown button, clearly indicated by an arrow or similar visual cue suggesting multiple options are available. This visual change is the first confirmation that the underlying functionality has been updated.

Following this observation, the user should then test both paths. By clicking the main, primary area of the button (which should remain the default action), they should confirm navigation to the chart creation page. This verifies that the existing, familiar workflow is still intact and functions correctly.

Subsequently, to test the new functionality, the user should repeat steps 2 and 3 (selecting a database/table and observing the dropdown button). This time, instead of clicking the main button, they should click the dropdown arrow to reveal the available options. From the presented menu, the user should select the option explicitly labeled as "Create dataset only" or a similarly unambiguous phrase. The expected outcome of this selection is a redirection to the dataset list page (typically /tablemodelview/list/), rather than the chart creation page. Finally, as a crucial validation step, the user must confirm that the dataset was indeed created successfully and is visible in the newly loaded dataset list. This comprehensive set of tests ensures that the new option functions correctly, the existing default remains operational, and the overall user experience is improved by offering greater choice and reducing unnecessary friction.

Enhancing User Experience in Superset Data Management

This proposed enhancement for Superset directly tackles a common usability issue, aiming to significantly improve the user experience for a broad range of users. By introducing the ability to create a dataset without being automatically funneled into the chart creation process, Superset becomes more accommodating to diverse data management strategies. For data analysts who might be preparing multiple datasets for a later comprehensive analysis, or for administrators setting up data sources for batch processing, the current forced redirection is a needless interruption. The proposed solution, a simple dropdown on the 'Create' button, offers a clear choice: proceed to visualize immediately or simply save the dataset and return to the list. This minimal change provides maximum flexibility, respecting the user's immediate intent.

Moreover, this update aligns with the broader goal of making Superset an even more intuitive and user-friendly platform. Reducing unnecessary clicks and cognitive load, especially in repetitive tasks, contributes to a more efficient and less frustrating workflow. The current behavior, while perhaps designed with the most common use case in mind, inadvertently penalizes users with different, yet equally valid, objectives. The acceptance criteria, meticulously detailed, ensure that this new functionality is not only implemented but also robustly tested, guaranteeing that existing workflows remain unaffected while new possibilities are unlocked. The verification steps provide a clear path for testing, ensuring confidence in the deployed solution. Ultimately, this feature empowers users by giving them agency over their workflow, making data preparation in Superset a smoother and more adaptable process. This kind of thoughtful iteration is key to maintaining Superset's position as a leading data exploration tool.

Conclusion: A More Flexible Superset for All Users

In conclusion, the presented enhancement to Superset’s dataset creation workflow promises a more intuitive and adaptable user experience. By transforming the singular 'Create' button into a strategic choice between creating a dataset and immediately proceeding to chart creation, or simply creating the dataset and returning to the list, we empower users with greater control. This addresses a genuine point of friction identified by the community, particularly for those who utilize datasets for purposes other than immediate visualization. The implementation, guided by clear acceptance criteria and a straightforward verification process, ensures that this feature enhances the platform without disrupting existing user habits. This improvement signifies Superset's ongoing commitment to evolving based on user needs, making powerful data exploration accessible and efficient for everyone, from novice analysts to seasoned data engineers.

For further insights into data visualization best practices and advanced Superset features, you can explore resources from Apache Superset itself or consult guides from organizations like DataCamp which often provide comprehensive tutorials and best practices for data analysis tools. For those interested in the technical aspects of open-source contributions and software development, GitHub is an invaluable platform for understanding how such features are developed and implemented.