FOSS4G 2025 — Notes From the Road

// 2025-09-15 · 3 min read

This is not an academic summary. Just some personal notes from a week at FOSS4G 2025 — what I saw, what resonated, and what I’m still thinking about.

The Big Picture: Product-Driven Innovation

This year’s conference felt very product-driven. A lot of the innovation is happening in the web space — platforms like ctrees, digital twins, and Re:Earth stood out. Cloud-native workflows were the central theme, and the momentum is clearly moving toward web-based products rather than desktop applications.

On the desktop side, QGIS still dominates, and QField Cloud was mentioned often as an important companion.

Rust Came Up More Than I Expected

CQL2 work, as well as several competing Zarr and GeoArrow implementations amongst web map tilers, some backend at RE:Earth, and smaller projects are being built in Rust. I didn’t expect that many mentions in a single week.

The COG vs Zarr Question

The classic discussion appeared everywhere. The consensus remains that they serve different roles rather than competing directly.

Open Data Cube: Pride and Frustration

Open Data Cube (ODC) is still a major focus internationally. Many groups are working on things very similar to what I’ve been building myself. It’s a mix of pride — because I clearly was not off-track with the general idea — and frustration — because I’m doing on my own what whole teams are doing elsewhere.

It also makes me reflect on the long-term sustainability of my approach. I’m not sure how much longer I’ll have the energy to keep pushing things forward on my own if no one else comes on board.

JRSRP Is Well-Respected

A positive surprise: JRSRP carries weight. I noticed that some DEA participants initially didn’t pay much attention to what I was saying, until I mentioned I work with JRSRP, at which point the conversations shifted noticeably.

RSS Is Still Well Aligned With Current Practices

From a technical perspective, my work is still aligned with current practices. For my own work specifically, I sometimes feel either ahead of the curve or simply following a different one.

The Reproducible Workflows Round Table Resonated

My round table on Nextflow + Nix for reproducible geospatial workflows resonated well. Many attendees struggle with exactly the same issues we do, but rarely speak about them openly. That session felt like the most honest conversation of the week.

The Rust Workshop

The workshop ran mostly as expected. Participation was low-ish (9 participants) and macOS support was a bit rough, but the content held up.

In hindsight, presenting in multiple formats — full talk, workshop, and lightning talk — would have reached different audiences more effectively.

Interest in My Work Wasn’t as Strong as Expected

I made a few solid contacts, particularly from the Pacific Islands. They’re just starting the journey and are not yet attached to any technology — which is a good position to meet people from.

One clear lesson: people engage more when there are tangible examples. Blog posts, runnable code, public repositories, tutorials. I should work on that.

Tangibility Changes Everything

To make my work more visible, I should probably do a proper series of posts and set things up so people can easily play with the code — maybe via GitHub Codespace.

On the same topic, I also need to break out of the “HPC silo” a bit. Not the system, but the data. There’s a lot happening in the open geospatial community, and being more outward-facing would help connect better with that world.

Convenience vs Convince

A recurring theme for me was the tension between convenience and convince. The broader community overwhelmingly defaults to Python, and I often find myself swimming against that current. It’s a good reminder to balance technical direction with what makes adoption easier. Perhaps I should wrap the library in some Python.

What’s Next

A few things I’m taking away from this week:

Thanks to JRSRP for supporting my attendance. It was a valuable week.

Type to search...