September 2025 Summaries

The blog post delves into two methods for managing and retrieving filter states when embedding a Superset dashboard into applications, emphasizing the importance of knowing the current state of the dashboard for purposes like saving configurations or syncing with other app components. The first method involves auto-emitted dataMask changes, where events are triggered automatically whenever there is a change in filter state, providing real-time data that can be useful for live analytics but may result in excessive data if users frequently adjust filters. The second method, using the getDataMask function, allows on-demand retrieval of the current filter state, suitable for saving specific dashboard states, syncing across dashboards, or generating shareable links. While observeDataMask is beneficial for continuous tracking, getDataMask is ideal for capturing a snapshot at a particular moment. By combining these approaches with the ability to set filter states, users gain comprehensive control over dashboard interactions within their applications, enhancing functionality through seamless integration.

The text explores the author's experiences with AI-assisted coding using the Claude Code model, highlighting the unexpected productivity boost and the realization that AI models are highly suggestible and guided by subtle signals from users. Despite working in open-source analytics for a decade, the author finds their coding speed accelerated due to AI, sparking an investigation into why their results were consistently superior to others using the same tools. Through the analysis of thousands of prompts, the author identifies that the success lies not in the complexity of prompts but in the micro-corrections made during interactions with AI, which are informed by deep domain knowledge and intuition. The text also discusses the significant advantage open-source projects have due to extensive training data available to AI models, unlike proprietary codebases. The narrative emphasizes the importance of context density and intuitive steering in maximizing AI's effectiveness, suggesting that AI acts as an expertise amplifier, enhancing the capabilities of developers familiar with the codebase.

Preset has introduced the Interactive Pivot Table, an enhancement to their data exploration tools, which builds upon the previously launched Interactive Table by incorporating similar interactive features into pivot tables. Powered by AG Grid Enterprise, this new tool allows users to sort, filter, and rearrange fields within the pivot table with a simple drag-and-drop interface, mimicking the ease of use found in spreadsheets while leveraging Preset's advanced capabilities. Users can also collapse and expand row groups, pin columns for better visibility, and switch value aggregations between sums, averages, and counts, enhancing the data exploration experience. The Interactive Pivot Table is designed to be more intuitive and flexible, providing users and teams with powerful tools for dynamic data analysis directly within dashboards, without needing to modify existing data charts extensively. Current Preset users can upgrade to this feature through the Chart Builder, and it is available to all users who sign up for Preset.