Replies: 3 comments
|
Also we'll cover https://discourse.mc-stan.org/t/cmdstan-stan-2-39-release-candidates/41122 as part of the roadmap when it the 2.39 release happens. |
0 replies
|
I'm also thinking about doing a TidyBayesPlot version or backend for BayesPlot.jl |
0 replies
|
Also something we should consider is how do we implement some form of consistency with the other CmdStan-based interfaces in terms of Parity. |
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
As the StanJulia ecosystem continues to mature, I wanted to open a broader discussion around priorities and direction for 2026 and beyond. The primary focus moving forward is to continue building a modular, federated, and maintainable Julia interface ecosystem around Stan and CmdStan.
Immediate Priorities
The next major priority is the development of StanLaplace.jl.
The goal of this package is to fully capture and expose the Laplace functionality available in CmdStan while following the broader StanJulia architectural direction of:
Federated package design
Dedicated command-line interfaces by functionality
Separation of concerns across the ecosystem
Lightweight and composable Julia packages
This package will serve as the dedicated Laplace interface within the StanJulia ecosystem and should align with the overall CmdLine-oriented architecture we have been moving toward.
CmdStan Alignment
2. Targeting CmdStan 2.38
Following the initial StanLaplace.jl work, the next major milestone will be targeting support for the CmdStan 2.38 release.
This will include:
Compatibility updates
API alignment
Feature exposure
Testing and CI improvements
Ensuring interoperability across the StanJulia packages
Repository Stabilization and Issue Resolution
3. Addressing Outstanding Repository Issues
After CmdStan 2.38 support is stabilized, we will prioritize working through the remaining filed issues across the StanJulia repositories.
This includes:
Bug fixes
Architectural cleanup
Documentation improvements
CI/CD stabilization
Cross-package consistency
Dependency modernization
The goal here is ecosystem stabilization and long-term maintainability.
Expanding the Stan Ecosystem in Julia
4. Follow-up Packages
Following stabilization efforts, attention will shift toward continued development and modernization of:
StanArm.jl
BayesPlot.jl
Loo.jl
Projpred.jl
These packages represent important parts of the broader Bayesian workflow ecosystem and deserve dedicated focus moving forward.
Future Ecosystem Expansion
5. Additional Standalone Julia Packages
Beyond the existing roadmap items, we also want to explore additional standalone packages that make sense for the Julia + Stan ecosystem.
Potential directions include:
Diagnostics tooling
Advanced posterior analysis
Variational interfaces
Simulation-based calibration tooling
Probabilistic workflow orchestration
ONNX/export integrations
Better interoperability with Julia probabilistic ecosystems
The broader vision is to continue making Julia a first-class environment for the Stan ecosystem while preserving modularity, composability, and maintainability.
I’d love feedback, ideas, and contributions from the community regarding priorities, architecture, and potential future packages.
All reactions