Julia Implementation project discussion for GSOC 2026

[GREENRAT-K405]

Hello @pradeeban @Rahuljagwani @mayureshkothare , I am moving towards finishing my GSOC 2026 proposal along with a ready prototype. I have around 50 impactful PRs merged throughout the control-core project in last 3-4 months along with thorough communication with mentors on daily basis.

I have few doubts regarding the project:

1. Do we need shared-memory in the julia version (currently, it is working only for linux OS that too implemented for C++ only).
2. Any particular preference for states in julia version? (python version is purely module based, c++ is class based). I have tried to leverage the benefits from both the concepts python version's module-as-singleton way and strict type-safety offerend in c++ version.
3. the write() and read() in julia causes conflicts because JIT confuses it for Base.write/read(), but I have appropriately renamed them in julia version.
4. How dynamic do we need the parsing in julia version to be? the python version uses literal_eval for it's dynamic nature to be able to parse so many different types of object types, c++ only parses flat numeric lists. I am planning on using JSON for parsing in julia, but if we need such dynamic range like python, we have to do it manually (I have tried it, it's just ~150 lines of extra code). If you have any alternative in mind please let me know.
5. Since the advanced implementation of julia are left open to us contributors I am mentioning them as "advanced delieverables" that are subject to change after meetings with mentors, (although I have already tried to include some of them in my project approach).

Thank you, for all your guidance and mentorship for the community.

[hxrshxz]

hey @GREENRAT-K405 I think GSoC isn't a competition or a sprint to rack up PRs, or making loud presence it's about genuine learning and collaboration. The PR and proposal speak for themselves. Good luck to everyone applying!
😃

[GREENRAT-K405]

hey @GREENRAT-K405 I think GSoC isn't a competition or a sprint to rack up PRs, or making loud presence it's about genuine learning and collaboration. The PR and proposal speak for themselves. Good luck to everyone applying! 😃

Good to know you understand the spirit of open source, I discussed same exact point not too long ago, If you need any help or would like to discuss any idea please let me know I am always happy to collaborate with others.

I wrote the way I wrote because I take time every month end to kind of give a progress report to mentors, I have been contributing to the project for some time now, I have no need or intention to boast anything about my contributions or my skills, my mentors will decide how good I am as they have been guiding me till now, all I know is I am here for a long run and not just for competition. 😁

[pradeeban]

1. Start with the core concore functionality and then build up from there with shared-memory and everything else. So, that follows a natural progression.
2. That is up to the contributor. We do not have any specific preference. You can justify your approach in your proposal.
3. great.
4. That is a bit too much of implementation-specifics. You can detail in your proposal and the actual discussions on implementation could wait until the GSoC period itself. We do not have to make that distinction and decision yet.
5. Yes, core deliverables followed by "if time permits."

[GREENRAT-K405]

Hello @pradeeban @Rahuljagwani @mayureshkothare , hope you are doing really well,

I am excited to say that I have completed both my proposal and prototype,
I took the last 5-7 days to focus on completing this. I hope to receive any and all the constructive criticism from your side, also if you would Like to see something added or removed do let me know.

I will email and cc the proposal shortly,
Also I will be having mid-semester exams for next 5 days but will try to contribute whenever required.

Thanks a lot for your time and activeness towards the org!!

[pradeeban]

awesome, @GREENRAT-K405. I have reviewed your proposal. Good work.

[GREENRAT-K405]

@pradeeban @Rahuljagwani @mayureshkothare
I have emailed my proposal, hope it finds you well.

Also, would be curious to know if we have any preferred way for measurements and benchmarking, for my proposal, I have used benchmarking similar how it was done for ZeroMQ implementation.

[GREENRAT-K405]

@pradeeban @Rahuljagwani @mayureshkothare
As of right now the Julia nodes and workflow almost always perform 6-10% better in terms of Lower latency and throughput even without any kind of optimization. (Tried in zeromq performance and file based performance measurements).

Should I focus on developing other extra samples to show higher performance in proposal?
(I have already sampled most of the nodes and workflow available in concore to my prototype).

I have included very realistic benchmark score which may or may not be very impressive.

Also should I share my prototype in email? (3-4 mb zip file)

[pradeeban]

@GREENRAT-K405 maybe unnecessary to email the prototype at this point.

I think, given you have already established your competence in coding, additional samples is not necessary as far as the GSoC proposal is concerned. But you are free to add those if you think they will help the project (rather than your proposal).