Question: Any plans to support probabilistic inference?

[Rot127]

We currently look for logic programming languages to build our binary analysis chain on.
But one requirement for us is, that it supports probabilistic inference.

Are there any plans to add an extension like that in the future?
E.g.: https://arxiv.org/abs/2508.15166

[justinjoy]

Thanks for the question. There are currently no plans to add probabilistic inference to wirelog's core.

wirelog is intentionally scoped as a small, deterministic, embeddable C11 incremental Datalog engine. Probabilistic inference, especially with correlated inputs as in the Praline paper, is a much larger semantic and implementation commitment involving provenance, constraints, solver/refinement policy, and numerical correctness.

That said, this may be a better fit for an application layer on top of wirelog. For example, wyrelog could potentially use wirelog as the deterministic Datalog substrate and implement domain-specific probabilistic or risk analysis above it.

The important question is the concrete use case and test target. Could you share more about what kind of probabilistic inference you need in your binary-analysis chain? In particular:

• What are the uncertain inputs?
• Are input facts independent, or do you need correlated-input bounds?
• What output do you need: marginal probabilities, lower/upper bounds, ranking, confidence scores, or threshold decisions?
• Do you have a representative benchmark or test case we could evaluate against?

With that information, we can judge whether this belongs in an external tool, an application layer such as wyrelog, or a small export/provenance hook in wirelog.