In the dev meeting today, folks suggested grabbing the code that's being used in the dashboard and looking at creating an LLM summary of the dev mailing list to use as a starting point. I extracted threads from May to now, and added in a previous report as a rough reference for tone/structure etc. I won't paste any of this into the document as I'm unsure how useful it is - if other folks could give some input here, that would be helpful.
# Project Activity: Selected subproject updates: ## Community & Ecosystem The community is discussing a proposal to canonicalize the variant extension type across Arrow implementations. This would improve cross-language interoperability for mixed-type columns (e.g., from Parquet). Contributors are weighing whether Arrow should define and maintain a standard set of extension types going forward. We’ve also seen strong uptake of the Kapa.ai documentation assistant (~850 queries/month), which is helping users discover relevant docs and resolve common questions without needing to open issues or join mailing lists. There is renewed interest in community health tooling—such as dashboards tracking stale issues and unmaintained components—driven by a desire to improve contributor visibility and sustainability. ## GPU Support A new arrow-gpu backend has been proposed for donation. Built on the wgpu abstraction layer, it aims to provide vendor-neutral GPU acceleration across compute kernels. The community is evaluating how this could integrate into Arrow’s compute model without introducing unnecessary complexity. ## Feather Format Feather V1 support has been deprecated in the C++ implementation, in line with the project's goal of simplifying its scope. Contributors agreed to remove both write and read support, though there is some discussion around retaining read-only fallback for legacy data. ## Intervals and Standards Alignment The Iceberg community has proposed a new 128-bit ANSI SQL-compatible interval type. Arrow contributors are considering whether this should be adopted natively or handled as an extension type. This work is part of broader efforts to ensure Arrow remains compatible with evolving lakehouse and SQL engine standards. ## Skyhook Skyhook has been removed from the Arrow monorepo and moved to its own repository. This aligns with our strategy of offloading niche components to simplify the core project and reduce maintenance burden. ## CI & Infrastructure Thanks to newly granted AWS credits, we are expanding our CI and benchmarking infrastructure. This includes adding GPU runners, improving performance tracking, and supporting larger-memory workloads. These improvements are expected to improve release robustness and enable performance monitoring across languages. ## Benchmarks Several contributors are working to make benchmarking more visible and comparable across C++, Python, and R. This includes discussions around standardized benchmark suites and dashboards to detect regressions earlier—motivated in part by recent performance issues observed in Arrow R 19.x and 20.x. On Wed, 2 Jul 2025 at 16:50, Neal Richardson <neal.p.richard...@gmail.com> wrote: > Hi all, > It's time for our quarterly ASF board update again. I haven't gotten the > reminder email yet but know it is coming (whenever the July meeting is), > and to give everyone enough time to provide input on the document, I wanted > to get started. Please have a look at [1] and add any updates you'd like to > include. I've filled in the basic facts to get us started. > > Thanks, > Neal > > [1]: > > https://docs.google.com/document/d/1fSYEmfGtmzjt1prHkdRoyxwGMmcX3onAStWArfiSBis/edit?tab=t.0 >
