Repository navigation
fix(runtime): preserve ETS tables for configured heirs - #50
Merged
Merged
Conversation
mremond
force-pushed
the
fm/task-9d9adf9b
branch
from
November 15, 2026 09:50
8597074 to
1814618
Compare
mremond
force-pushed
the
fm/task-9d9adf9b
branch
from
November 15, 2026 12:22
1814618 to
1ddaedb
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Why
Configured ETS heirs lost their tables and contents when the owner exited.
What changed
Before and after
Before: Both native scheduler modes received EXIT instead of the required transfer notification in the baseline regression.
After: Twenty native and ten reference runs demonstrate inheritance; the full native repeat passes 502 tests with two external-module skips.
This changes runtime ownership and message delivery, not a graphical interface; native and reference transcripts retain the observed table state.
Checked
Live validation: Demonstrated for this version. 3 of 4 scenarios driven live in the product.
How it was checked
Built the runtime before the compiler with the pinned toolchain and locked dependencies.
Ran cargo clippy --workspace --all-targets --locked -- -D warnings and cargo test --workspace --locked --no-fail-fast; 2,196 Rust tests passed with one ignored measurement.
Ran ./target/release/nera test -j 1 --timeout 30 and the allocator, stacktrace and shutdown regression scripts.
The first full native suite observed a tcp_reactor_perf batch-variance failure; one full serial repeat passed using identical compiler, runtime and fixtures.
The original failure remains unexplained. Historical and repeat other-workload concurrency is unknown; no exclusive-host, host-load cause or flake claim is made.
The passing performance fixture prints no batch vectors or threshold margins. All 100 isolated runs on main and 100 on the change passed its thresholds, but did not reproduce the full-suite context.
All ten shutdown cases completed thirty samples; the largest median was 1.54 ms against the 5 ms limit, and the largest sample was 1.98 ms.
The shutdown gate checks medians, not maximum latency. The native baseline failed at the missing transfer while the unchanged reference fixture passed.
Risk
Risk above low: Blast radius, medium — Changes common process-exit cleanup and ETS ownership registry operations.; Novelty, medium — Adds owned heir gift storage beside existing table and message machinery.
Full risk assessment
Not done
Changing heirs with ets:setopts/2 remains unsupported; access-control behavior is unchanged.
Windows and target-native macOS x86_64/Linux arm64 execution were not performed.
The original TCP timing failure remains unexplained; concurrency and passing threshold margins remain unknown.
Pipeline
Updates from git push no-mistakes
Infinitary checked this version: an independent review that also judged the documents, and live validation the review accepted.