A topic to informally discuss Julia programming or related technical projects that we’ve been working on. Tell us about one of your recent projects, either at work or personal projects. What’s interesting about it? Are you having trouble with anything?
Great initiative! When it comes to Julia: Joovy.jl and my Flexible Julia plugin.
Flexible Julia is just a decent JetBrains IDE implementation for the language, with profiler, Jupyter/Pluto support, remote sessions, complete grammar reimplementation (on top of LS support) and various features too long to mention. Most recently I added an MCP server and Chat AI interface to make it easy to work with AI whilst there.
Joovy on the other hand is a project to optimize compilation times. It sits between the compiler and the IDE and tries to anticipate what to precompile when so that your actual runtime is faster. There are various methods surrounding it that I documented in another thread.
Lastly, I also work on & off on an In-IDE visual modeling plugin for Julia. It is sitting on my desk, but hard to complete without larger projects and some external feedback, so I only come back to it when I am in the right mood 
Although I haven’t written a lot of Julia code in the past few months, I finished writing the manual for a product my team built (and sold) which uses juliac. It uses simulations to help administrations and aerospace companies coordinate to avoid interference across spacecraft and ground stations. I’ve started work on that project at about the time the juliac repo was created. There’s a bunch of ETL, lots of domain-specific stuff, a Python server and an RDBMS. On the client side it’s an Electron app which bundles a C++ binary that does most of the business logic, and passes data through the C FFI to Julia for all the simulations. I’m using juliac to compile code to a shared lib, linked to the cpp bin.
On other news, we keep using our, mostly in Julia, simulators to do work for various space missions and adjacent projects. The Julia code is now flight-proven! The output of our Julia simulators matches real-world telemetry data from in-orbit satellites with high accuracy. We’ve used Julia to successfully design, validate and fly spacecraft, rescue a satellite while in-orbit and troubleshoot a wide variety of third-party hardware and software.
We were one of the two aerospace companies in my country that were selected to join the European Space Agency incubator program, and received additional rewards since we scored the highest in their evaluations. We’re raising some funds too.
Another space program in my country was recently announced, and we expect to use our Julia tools to pretty much back the majority, if not all, the upcoming proposals, plus directly support the space center here etc.
All thanks to Julia! 
I haven’t stated this, but I’m developing a package to utilize MANOVA models as well as posthoc tests for them. I’m also building a dyadic model for my research to model how languages migrate throughout space and time.
Well, I like optimization and manifolds, so I am mainly considered with Manifolds.jl and Manopt.jl – for quite a while actually.
I do like that it requires to rethink quite a few problems and properties when leaving Euclidean spaces.
A challenging part is, that several different people usually have different ideas and opinions what manifolds in code should be able to do – and sometimes even whether that is useful to have.
I’m giving PortfolioOptimisers.jl a systematic clanker assisted audit. Finally adding all the missing references, properly describing algorithms in docstrings, increasing coverage to as close to 100% as i can. Adding more reference tests, fixing bugs.
I’ve set up an automated self-healing CI process that runs JET.jl, CodeComplexity.jl and coverage reports on a weekly basis as well as ratchets that warn me when these quantities worsen on pushes.
I have to say Opus 5 is phenomenal for this kind of work. I’m really pleased.
I’ve been thinking of doing a BAM view package to make it easier to do downstream analysis on BAM files. Given a BAM file one can apply different types of views to it and get the information needed out of it, a bit like samtools view / mosdepth but more configurable, handling paired-end reads, UMIs, etc.
I’ve been following Reactant and KernelAbstractions. Finally took some time and build a differentiable, finite volume MHD solver with the help of ChatGPT. I’m really amazed by how well thought out the packages are and what they enable people to build.
I’ve been exploring the possibility of implementing functionality to create native Excel charts in XLSX.jl. With Claude’s help, I mapped out a 5-stage plan and I’m currently part way through stage 2. To be honest, there is no way I’d have been able to begin this without Claude’s support. There is a lot of complexity in Excel’s use of XML to describe everything, and Claude can assimilate this in a way I don’t think I could.
Hopefully, we’ll produce something worthwhile in the end…
I’m currently working as a spare time project (and with AI assistance) on Home · SymbolicUncertainties.jl
A Julia package for purely symbolic propagation of measurement uncertainties, implementing the methodology of JCGM 100:2008 (GUM - Guide to the Expression of Uncertainty in Measurement)
It would be really neat to have LibreOffice odt file support. Even though the name of the package is specific to Excel.
Vibe-coding a remesher for fun :
It take a messy, high resolution mesh (from a scan, sculpt or generative model) and tries to remesh it with a lower resolution, well behaved mesh.
It works as well as open-source tools (Instant Mesh, AutoRemesher) from the few tests I did. But it’s very slow. I’m not sure what the SOTA is in the domain but you need to add-in some prior knowledge about anatomy, animation, etc to do a good job, so a machine-learning part is probably necessary.
