Uses a TLS-detecting listener that peeks the first byte of each
connection. TLS handshakes (0x16) proceed normally; plain HTTP
requests get a 301 redirect to the HTTPS URL. This handles
http://localhost:3000 gracefully instead of showing a cryptic error.
Move EBS, cluster, and node message types from config/ to types/.
Remove duplicate NodeDiscoverResponse from gateway and daemon.
Delete dead EC2StartInstances types and tests from config.
Update all imports across source and test files.
Squash of dev commits: gateway core routing/policy/middleware tests,
IAM auth and handler tests, awserrors tests, respondWithJSON helper,
utils refactor into focused files, and daemon test speedups.
Drop the external gziphandler dependency in hiveui. The replacement
uses a simple http.ResponseWriter wrapper that checks Content-Type
against an allowlist before enabling gzip compression. Same behavior,
zero external deps.
Add awserrors.ValidErrorCode() to validate error strings against
ErrorLookup before sending to clients. Replace hardcoded
"ValidationError" strings in utils.go with awserrors constants.
The MIN_LOC=100 threshold checked total diff lines, not instrumentable
lines. A commit with many non-instrumentable lines (structs, comments)
but only 1-2 instrumentable lines would unfairly fail at 0% coverage.
Add a minimum instrumentable lines threshold (5) so the check skips
when the sample size is too small to be meaningful.
During cluster restart, restoreInstances resets running instances (with
dead QEMU) to pending for relaunch, but didn't reset LaunchTime. The
pending watchdog (5min timeout) saw the stale LaunchTime from the
original launch and immediately marked instances as failed/terminated,
preventing any relaunch from succeeding.
Extract duplicated respondWithError closures into a single package-level
function. Guard against nil ImageId in handleEC2RunInstances to prevent
panic under race conditions. Fix diff-coverage base ref for main branch.
Julian Sommer