Skip to content

docs: v0.7.0 Track A commit-2 design (hierarchical IRQ-aware RTA)#146

Merged
avrabe merged 1 commit intomainfrom
docs/v0.7.0-hierarchical-rta-design
Apr 24, 2026
Merged

docs: v0.7.0 Track A commit-2 design (hierarchical IRQ-aware RTA)#146
avrabe merged 1 commit intomainfrom
docs/v0.7.0-hierarchical-rta-design

Conversation

@avrabe
Copy link
Copy Markdown
Contributor

@avrabe avrabe commented Apr 23, 2026

Summary

Design doc for the next v0.7.0 Track A commit. Ready to hand off to an agent when PR #145 merges.

What's in

  • Current rta.rs walkthrough with line refs (line 34 RtaAnalysis, line 121 the fixed-point call, property accessors at 195/214/240)
  • Gap analysis: no ISR tier, no jitter use, BCET/WCET collapsed, no probe awareness
  • Two-tier algorithm pseudocode (ISR tier steals capacity → task RTA on residual + jitter + ISR interference)
  • Five new diagnostics with severity + emission criteria
  • File-by-file change scope (~700 LOC across rta.rs, property_accessors.rs, scheduling_verified.rs, diagnostics.rs)
  • 10 unit tests + 3 fixture tests with expected-output JSON, including the critical non-regression test
  • Out-of-scope list (PIP/PCP → v0.7.1; cache interference → v1.0+; MBPTA → v1.0+)
  • Commit message template + agent-executable implementation brief

Why this exists

v0.7.0 Track A commit 2 is the big one — the actual algorithm change. Landing the design doc first means:

  1. You can review the approach before ~700 LOC of code hits review
  2. The agent that implements it gets a complete brief, reducing briefing-time errors
  3. Future contributors have a rationale artifact, not just a diff

Non-regression is the critical gate

Any model without Spar_Timing::* properties must produce byte-identical RTA output to today. This is explicitly called out as the first test to run.

Not yet

  • No code changes in this PR
  • Lean convergence sketch for the jittered recurrence ships as commit 3 (proofs/Proofs/Scheduling/RTAJittered.lean)
  • Integration tests + COMPLIANCE.md final wording ship as commit 4

Test plan

  • Review the algorithm (tier split, jitter fold, BCET/WCET handling)
  • Confirm the 10 unit-test cases cover the behavior space you care about
  • Confirm the out-of-scope list matches your priorities
  • Agree the commit-message template + agent brief are ready to execute

🤖 Generated with Claude Code

@codecov
Copy link
Copy Markdown

codecov Bot commented Apr 23, 2026

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

Captures the design for the next Track A commit after PR #145 lands.
Two-tier analysis: ISR layer preempts all tasks, residual capacity
feeds classical task-priority RTA; Tindell-Clark jitter terms fold
into the fixed-point recurrence; Compute_Execution_Time's Time_Range
consumed as (BCET, WCET) with a new ResponseBand diagnostic.

Includes: current rta.rs walkthrough with line refs, gap analysis,
algorithm pseudocode, new diagnostics table, file-by-file change
scope with LOC estimates, 10-unit-test plan + 3 fixture tests,
out-of-scope list (PIP/PCP, multi-processor ISR migration), commit
message template, and an executable agent brief ready to hand off
when #145 merges.

Non-regression is the critical gate: models with no Spar_Timing::*
properties must produce byte-identical RTA output.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
@avrabe avrabe force-pushed the docs/v0.7.0-hierarchical-rta-design branch from fab221c to 5067481 Compare April 24, 2026 03:50
@avrabe avrabe merged commit c5b1ac2 into main Apr 24, 2026
11 checks passed
@avrabe avrabe deleted the docs/v0.7.0-hierarchical-rta-design branch April 24, 2026 20:36
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant