On Monday, I met with Patrick for another meeting to firm up some details about the geospatial DSL. After meeting with Patrick, I spent a while working on the first draft of my proposal based on the design goals we had discussed, and then sent him a first draft for review a day later. We got it to what I thought was a reasonably good place, although it definitely needed some polish.

Then one Wednesday, I met with Anil and Carl, my DoSs. They gave me some constructive criticism on the proposal, and in the end we decided it wasn’t the best route to go down, and so pivoted to the OpenStreetMap DSL! This project seems really interesting and actually addresses a problem I’ve had in my normal life. I think it will also have a good evaluation story and should hopefully do well. It was a bit hectic fleshing out the project on Wednesday night and then writing the proposal on Thursday, especially given term and lectures started then, but I ended up with a proposal I’m really happy with and excited to get started on. It also turns out that Anil can’t be my marking supervisor, but thankfully Jon Sterling agreed to.

Over the weekend, I’ve started looking into OSM a bit more, but haven’t done much else yet.

Transclude of part-ii-project-proposal

At the start of this week I still needed to decide whether I was doing the eBPF project or the geospatial DSL. My supervisor Patrick and I had another meeting on Wednesday, which was really useful for clarifying a bit more about what he had in mind for the geospatial DSL and where it could go.

I read the Yirgacheffe paper¹ (which was sort of where the original idea for the DSL came from) and played around with it a bit, and found it interesting. Given the greater opportunity the DSL provided, and the fact the eBPF project seemed like it might get caught up a bit in the writing of the backend of the compiler (lowering to eBPF bytecode or LLVM IR), I decided to go with the eBPF.

I don’t really have much experience at all working with geospatial data, but I’m excited to get started!