FE-1271: Fire predicate transitions in the same step their guard becomes true

[claude]

Requested by Chris Feijoo · Slack thread

🌟 What is the purpose of this PR?

Before: a predicate (boolean guard) transition whose guard was true could not fire in the very first simulation frame, and could never fire in two consecutive frames — a persistently-true guard fired only every other frame (every 2·dt), even though the docs say a true guard "fires immediately".

After: a true guard fires in the same simulation step it becomes true, including the first frame and every consecutive frame while it stays true. Stochastic-rate transitions behave exactly as before.

How: boolean guard results were converted to a stochastic rate of Infinity and pushed through the exp(-rate * timeSinceLastFiringMs) <= U1 firing test. timeSinceLastFiringMs is 0 on the first frame and resets to 0 on firing, and Infinity * 0 = NaN silently fails that test. Predicate results now bypass the stochastic test and fire directly when the guard is true; the numeric-rate path is untouched.

Fixes FE-1271

🔗 Related links

• Slack thread reporting the behavior (internal)

🔍 What does this change?

• compute-possible-transition.ts: boolean lambda results no longer map to an Infinity rate — a true guard fires directly, a false guard doesn't; only numeric rates go through the exp() test
• Regression unit tests in compute-possible-transition.test.ts: a true guard fires with timeSinceLastFiringMs = 0 (the case the test helper's 1.0 default was masking); a false guard never fires
• Regression test in compute-next-frame.test.ts: a persistently-true predicate transition fires on every frame from frame 1 onwards
• Changeset for @hashintel/petrinaut-core (patch)

Pre-Merge Checklist 🚀

🚢 Has this modified a publishable library?

This PR:

• modifies an npm-publishable library and I have added a changeset file(s)

📜 Does this require a change to the docs?

The changes in this PR:

• are internal and do not require a docs change
  • the user docs already describe the fixed behavior ("fires immediately"); this makes the engine match them

🕸️ Does this require a change to the Turbo Graph?

The changes in this PR:

• do not affect the execution graph

🛡 What tests cover this?

• New: predicate firing at timeSinceLastFiringMs = 0 and consecutive-frame firing (see above)
• Existing: full @hashintel/petrinaut-core unit suite (839 tests) passes

❓ How to test this?

1. Checkout the branch
2. Build a net with a predicate transition whose guard is always true (e.g. a self-loop) and run the simulation
3. Confirm the transition fires on the very first frame and on every subsequent frame, rather than every other frame

---

Generated by Claude Code

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
hash	Ready	Preview, Comment	Jul 24, 2026 4:45pm
petrinaut	Ready	Preview, Comment	Jul 24, 2026 4:45pm

1 Skipped Deployment

Project	Deployment	Actions	Updated (UTC)
hashdotdesign-tokens	Ignored		Jul 24, 2026 4:45pm

[cursor]

PR Summary

Low Risk
Localized simulation-engine firing logic with targeted regression tests; stochastic transitions are untouched.

Overview
Predicate (boolean guard) transitions now fire in the same simulation step when the guard is true, including the first frame and every frame while the guard stays true. Previously, true guards were treated as an infinite stochastic rate and passed through exp(-rate × timeSinceLastFiringMs); with timeSinceLastFiringMs === 0 after a firing (or on frame one), that produced NaN and blocked firing, so persistently-true predicates only fired every other frame.

In compute-possible-transition.ts, boolean lambda results bypass the exponential firing test: a true guard fires immediately, a false guard does not. Numeric-rate (stochastic) transitions are unchanged.

Regression coverage was added in compute-possible-transition.test.ts (guard true at timeSinceLastFiringMs = 0, false guard never fires) and compute-next-frame.test.ts (always-true predicate fires on frames 1–4). A patch changeset was added for @hashintel/petrinaut-core.

^{Reviewed by Cursor Bugbot for commit fc90a42. Bugbot is set up for automated code reviews on this repo. Configure here.}