Conversation
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 7a92eaea1f
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
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".
|
|
||
| (entrypoint, rendered_env) | ||
| }; | ||
| let entrypoint = resolve_relative_entrypoint(entrypoint, component_source_dir.as_deref())?; |
There was a problem hiding this comment.
build_run_plan now unconditionally calls resolve_relative_entrypoint, which errors when the program starts with a relative path containing a separator (for example ./start.sh) and AMBER_COMPONENT_SOURCE_DIR is missing. That environment variable is only populated by the direct runner path in cli/src/main.rs, while non-direct helper launches (for example the HelperRunner path in compiler/src/targets/mesh/docker_compose/mod.rs) only pass AMBER_DIRECT_ENTRYPOINT_B64/template/mount env vars. As a result, existing docker-compose/kubernetes workloads that require helper mode (runtime templating or mounts) and use a relative image entrypoint will now fail at runtime.
Useful? React with 👍 / 👎.
1 participant
This PR adds a new "direct" backend which runs programs natively but with sandbox like bubblewrap (linux) or seatbelt (macos). The way to use this is
amber compile --direct ...which will output an artifact that can then be run withamber runwhich acts as a sort of init process that manages the process tree. The mesh/capability semantics otherwise are preserved.