Google Summer of Code: 0x02

we start with some raw code contribution

Disclaimer: the below process is not a roadmap for GSoC. It is just a document of my own experience which I want to share with anyone interested.

Contributor phase

I already knew about LabLua organization because one of my batchmates had participated in GSoC in 2024 ¹. Naturally curious I tried understanding the codebase. I was able to build the project and run the example scripts, and I got interested. I checked open issues and saw one related to LLDP example.

I was able to get my first PR merged in the codebase on Dec 29, 2025, 7 days after I had created the PR. The review process was rigorous, strict concerns over code hygene, commit history and documentation. I was impressed by the amount of things I learnt apart from technicalities of LLDP.

Member phase

After 6 successful PRs merged, I was invited to join the GitHub organization on 17 Janurary, 2026. I did not stop there, I contributed more, and it came naturally, as the issues I was working on, kind of helped me see other issues with the codebase. One thing led to the other and I was touching core Lunatik object and class code. I made some improvements, specifically with latency.

2 PRs I had lot of fun while writing in this period are the following:

• wrap class methods eagerly instead of via __index #410
• add support for non-shared objects in shared class #434

Proposal phase

Google announced LabLua was going to participate, as I was fairly acive on the project Matrix, I kept regularly checking project ideas page. The proposal which I think I could do was Lunatik Bindng for sched_ext.

As a next step I asked the maintainers about this on the very day the mentoring organizatons were announced (specifically February 19, 2026). The maintainer pushed me to take on something more challenging, they said:

I’m afraid, due to your current level of knowledge on Lunatik, sched-ext could be a bit easy to you.. we expect that collaborators have adequate challenging level..

And I at once, started to reason what could be a nice proposal. I merged two ideas mentioned on the page and came up with my own view. I created multiple gists to share my ideas, and the biggest feedback I got was:

I think it’s important that you dig into the foundations of this work.. you have walked a good way on the “how’s”.. but you need the “why’s” (but I see you are already looking for it as from your TC doc)

And there I was, trying to find a solid why to my proposal. Because when you only know a hammer.. everything is a nail.. it shouldn’t be the case here.. Lua is pretty much a glue language and I needed to treat it like that.

Proposal

My final proposal which I submitted after multiple review iterations is here.

I proposed to add Lunatik Bindings for TC, Sched, and eBPF maps. The project promises to make Lunatik work with eBPF and not against it. If I have to summarize my project, I’d say:

eBPF defines the structure, fast path. Lua provides the dynamic policy, slow path.

GSoC results

On April 30, 2026 the accepted projects were announced and at 11:54 PM IST, I recieved a mail from GSoC and I was selected ².