You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This commit was created on GitHub.com and signed with GitHub’s verified signature.
Improvements
Sandbox run and exec now wait for execution completion and stream consumption in parallel, eliminating tail latency from sequential polling for exit codes — Jeff Haynie#1416
Sandbox run and exec use server-side long-poll for execution status instead of client-side polling loops (previously up to 15 retries × 1s) — Jeff Haynie#1416
Sandbox cp directory uploads now use tar.gz archive transfer instead of reading all files into memory individually — Jeff Haynie#1416
Sandbox cp directory downloads now use tar.gz archive transfer instead of running find and downloading each file separately inside the sandbox — Jeff Haynie#1416
Sandbox file downloads use streaming pipelines instead of buffering entire file contents in memory — Jeff Haynie#1416
Sandbox CLI commands use cache-first lookup for routing context (region and org), reducing redundant API resolve calls — Jeff Haynie#1416
Resource cache SQLite database now uses WAL journal mode and busy timeout for improved concurrent access — Jeff Haynie#1416
Sandbox exec stream grace period reduced from 5s to 500ms after execution completes — Jeff Haynie#1416
Bug Fixes
Resource cache errors no longer cause sandbox CLI commands to fail; errors are caught and treated as cache misses — Jeff Haynie#1416
Sandbox get now handles null org field in responses from the server's pending cache, preventing validation errors on newly created sandboxes — Jeff Haynie#1416
Relax stream namespace timestamp fields to be nullable, preventing validation failures when the server returns null timestamps — Jeff Haynie#1406