Repository navigation
Add a debug check that user actions do not delete secondaries - #711
Conversation
|
@codex review |
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
|
Codex Review: Something went wrong. Try again later by commenting “@codex review”. ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
If Codex has suggestions, it will comment; otherwise it will react with 👍. Codex can also answer questions or update the PR. Try commenting "@codex address that feedback". |
|
/run-test |
|
As the automatic AI review is broken: manual AI review with Codex 6.1 SoL ultra found no issues with the PR. |
…lone The secondaries attached to a replayed GPU step are placement-new'd pool objects that are already transported on the GPU. A user stepping action that removes, replaces, kills or reweights them desynchronizes the CPU replay from the GPU transport, and one that deletes a removed secondary corrupts the replay storage. Such decisions have to be taken by the stepping action on the GPU before a secondary is created (ClassifySecondary). Snapshot the GPU-born secondaries before the user stepping action and throw a descriptive exception if their identity, status or weight changed. This is purely an extra consistency check for debugging such mismatches: it is compiled out in release builds (NDEBUG, like the asserts in this file), changes no behavior for compliant user actions, and costs nothing in production. Assisted-by: Claude Code (Claude Fable 5)
2c98ac3 to
cfdc2be
Compare
|
/run-test |
The secondaries attached to a replayed GPU step belong to particles that are already transported on the GPU. A user stepping action that removes, replaces, kills or reweights them introduces a difference for the CPU replay from the GPU transport, and one that deletes a removed secondary corrupts the replay storage. Such decisions have to be taken by the stepping action on the GPU before a secondary is created.
As this is a non-trivial check, it is for debug only, as this should normally never trigger.
Assisted-by: Claude Code (Claude Fable 5)