mirror of
https://github.com/paperclipai/paperclip.git
synced 2026-10-25 17:45:53 +01:00
Copy the repository .cargo directory into the Daytona runner Docker build stage, so the production cargo build picks up the raised retry count and transfer timeout instead of the cargo defaults. Add .cargo to the audited Daytona image input paths, so a future change to that file invalidates the cached image. Make Arm B of the network stall differential harness unset CARGO_NET_RETRY and CARGO_HTTP_TIMEOUT before it runs cargo fetch, so the harness tests the repository config instead of a value the caller happens to have exported. Correct the .cargo/config.toml comment: the retry count rises from Cargo's default of 3 to 5, which is two more retries, not one. Co-authored-by: Paperclip <noreply@paperclip.ing>