A monthly newsletter, mostly on Julia internals: a biased, incomplete, editorialized list, with contributions from the community. Subscribe by email on Discourse.
The deadline for submitting a proposal to JuliaCon 2025 is extended to Feb 14th.
Pkg now supports 'apps' for running programs directly from the terminal without invoking Julia, using isolated environments for each app; more details are available in the PR and slack discussion.
On a discussion about introducing Rust alongside C in Julia’s source code, Stefan provided interesting insight into what Julia looks like as a program: code generation is done mostly by C++ because it is the main interface for LLVM, and C is used for OS-interfacing, significantly limiting the utility Rust would provide.
PR #57049 introduces LLVM-based feature detection for CPU parsing.
The difficulty of writing multithreaded code, and footguns related to the difference between thread and task, the danger of moving tasks between threads, and the subleties of properly setting up thread- or task-local storage come up often. This is a recent good discussion of the topic. It reminds me also of the recently merged perprocess / perthread / pertask constructs which help manage some of these subtleties.
An interesting suggestion for having julia automatically logging to the terminal to current state of progress through a script.
Unicode v16 is now supported in Julia source files. We get access to the new "face with bags under eyes" emoji but Unicode still doesn't have superscript “q”.
The Center for Quantum Networks is running many bounties and minigrants on a variety of support packages for quantum information science.
SciML is running a large set of funded small projects – a great opportunity to contribute to the community and be paid for it.
ChainRules, a foundational package to the Julia autodiff effort, is looking for a new primary maintainer.
Speculator.jl is a fun new tool to speculative precompile functions (in the background) to reduce latency (also in interactive dev use).
Makie 0.22 is now out, with further improvements to geometry primitives and 3D capabilities.
BorrowChecker.jl is a new package that introduces Rust-like ownership and borrowing semantics to Julia, aiming to improve memory safety. Related to this earlier discourse discussion.
Statistics.median is now available for StaticArrays, providing efficient median computation via bitonic sorting.
AlgebraOfGraphics.jl (inspired by R’s grammar of graphics ggplot, built on Makie) is working surprisingly well with sophisticated edge-case composition of plots.
SatelliteToolbox.jl is officially v1 after 10 years – used in attitude and orbit control of actual satellites.
QuantumCircuitDraw.jl is a new package for visualizing quantum circuits in the style of Qiskit.
The newly published book The Computer Algebra System OSCAR: Algorithms and Examples (2025) delves into OSCAR’s role in algebra, geometry, and number theory, featuring expert-written code and interdisciplinary applications.
The experimental PR for Oscar.jl introducing Clifford algebras, orders, quadratic spaces, and lattices has been merged!
A new version of the book “Fundamentals of Numerical Computation” is out, with julia/python/matlab snippets for each algorithm.
Interesting discussion on techniques used for function approximation with polynomials.
PeriodicMatrices.jl was recently released to handle periodic time-varying matrices.
The JuMP Steering Committee announced a major grant from the Breakthrough Energy Foundation, through NumFOCUS, to enhance the performance of JuMP and HiGHS on open energy models.
JuMP and its SCIP solver enabled recent work on automated branching rule generation, solving a weighted minimum set covering problem with 15,782 sets in seconds!
ChainRules Thunks are now fully integrated into Zygote.jl, following a challenging but rewarding implementation process.
Effort.jl combines Turing.jl, SciML, and Lux.jl to train fast, accurate surrogate models for cosmological analysis, with applications to DESI and Euclid surveys.
This type of discussions come up every so often and occasionally have new insight: Autodiff in the Julia multiple-dispatch staged-compilation model vs the Python reimplement-the-universe model.
Clang gets LLVM-based tooling that checks at runtime and compile-time for realtime violations.
JAX Under the Hood explores how understanding JAX’s architecture enhances performance optimization, efficient coding, and profiling in scientific computing.
Chris Rackauckas discusses building products for open-source sustainability, highlighting the work of MIT Julia Lab, PumasAI, and JuliaHub, in a recent non-technical talk with OpenTeams.
See also: French community newsletter, community calendar, minutes of triage meetings
You can engage in the discussion of this newsletter on Discourse