This lightning talk will present the current state of tooling for running spatial computations on the client-side (eg. on a personal computer or phone, primarily within a browser context). It will ask the audience to re-consider what is possible on the client-side, and what this means for the architecture of geospatial applications.

In the 2000s and 2010s, web browsers were limited to HTML (for structure), CSS (for style), and JS (for interactivity). Spatial tooling in the browser largely consisted of Javascript frameworks such as Leaflet or Mapbox for rendering interactive maps. There were limited attempts to implement core geospatial processing functions (predicates, CRS reprojection, spatial statistics) in Javascript - turf.js being a notable exception. Rather, most geospatial processing was done in traditional server-side languages like C/C++, Python, or R. These programming languages developed powerful spatial processing ecosystems and libraries, including PROJ, GEOS, GDAL, GeoPandas, and more.

The web browser development community, mindful of the performance limitations of Javascript in the early 2010s, began looking for ways to bring native-level performance to browser. The first attempts were asm.js and NaCl, which ultimately became WebAssembly (Wasm). Wasm was supported by all major browsers in 2017 and became a W3C recommendation in 2019.

WebAssembly provides a way to run non-Javascript code within a web page. It does this by serving as a compilation target/ instruction set which other languages can be compiled to. This allows any program written in C/C++ to be compiled to WebAssembly and run on the client, including foundational geospatial libraries like PROJ, GEOS, and GDAL.

This new architecture opens up a plethora of new opportunities for geospatial software.

Security
- Because it can run on the client-side, private data does not need to be shared with a backend
- Imagine fitness software that gives you powerful analytics on GPS-based tracking data, without needing to transfer data to a backend.

Performance/ Scalability
- With a traditional backend handling data processing, the backend must scale with the number of users.
- With processing done on user devices, the users 'bring their own compute, which makes geospatial applications highly scalable and resilient.

User interface
- Results can be computed and shown to the user instantaneously, mirroring the experience of using desktop-based map applications.

WebAssembly is still relatively new, and is not well-known among the geospatial community. With this talk, I hope to change that a little bit. I am very excited by the possibilities that it holds for more secure, decentralized, and performant geospatial applications.