CHANGELOG.json

{
  "name": "@rushstack/rush-client-core",
  "entries": [
    {
      "version": "0.2.4",
      "tag": "@rushstack/rush-client-core_v0.2.4",
      "date": "Fri, 30 Oct 2026 21:34:20 GMT",
      "comments": {
        "dependency": [
          {
            "comment": "Updating dependency \"@rushstack/node-core-library\" to `5.25.2`"
          },
          {
            "comment": "Updating dependency \"@rushstack/rush-daemon-protocol\" to `0.5.3`"
          },
          {
            "comment": "Updating dependency \"@rushstack/rush-daemon-transport\" to `0.4.3`"
          },
          {
            "comment": "Updating dependency \"@rushstack/heft\" to `1.3.5`"
          }
        ]
      }
    },
    {
      "version": "0.2.3",
      "tag": "@rushstack/rush-client-core_v0.2.3",
      "date": "Tue, 27 Oct 2026 00:20:25 GMT",
      "comments": {
        "none": [
          {
            "comment": "Allow slow daemon startup in queued-request resend integration tests."
          }
        ],
        "dependency": [
          {
            "comment": "Updating dependency \"@rushstack/rush-daemon-protocol\" to `0.5.2`"
          },
          {
            "comment": "Updating dependency \"@rushstack/rush-daemon-transport\" to `0.4.2`"
          },
          {
            "comment": "Updating dependency \"@rushstack/heft\" to `1.3.4`"
          }
        ]
      }
    },
    {
      "version": "0.2.2",
      "tag": "@rushstack/rush-client-core_v0.2.2",
      "date": "Tue, 06 Oct 2026 00:04:37 GMT",
      "comments": {
        "patch": [
          {
            "comment": "Make daemon startup, restart and crash recovery reliable: concurrent clients share one startup and connect as soon as it is ready, requests follow daemon restarts, and a liveness ping detects an unresponsive daemon. A crashed daemon's stale files are removed, and on Linux the operations it left running are stopped. Errors say why no daemon answered. After a daemon fails to start, clients wait 15 seconds before launching it again. Add APIs for restart reasons, queue details and crash reclaim."
          },
          {
            "comment": "Fix the error reported for a crashed daemon, which included the stack of its crash report when FORCE_COLOR was set."
          }
        ],
        "dependency": [
          {
            "comment": "Updating dependency \"@rushstack/node-core-library\" to `5.25.1`"
          },
          {
            "comment": "Updating dependency \"@rushstack/rush-daemon-protocol\" to `0.5.1`"
          },
          {
            "comment": "Updating dependency \"@rushstack/rush-daemon-transport\" to `0.4.1`"
          },
          {
            "comment": "Updating dependency \"@rushstack/heft\" to `1.3.3`"
          }
        ]
      }
    },
    {
      "version": "0.2.1",
      "tag": "@rushstack/rush-client-core_v0.2.1",
      "date": "Wed, 30 Sep 2026 00:18:08 GMT",
      "comments": {
        "patch": [
          {
            "comment": "Make the daemon startup reservation bounded and verifiable: record owner and launcher PIDs, release it when the launcher exits before readiness, accept a ready daemon despite a reservation, reclaim stale reservations without waiting for the deadline, and include the launcher log's last error lines in startup failures."
          }
        ]
      }
    },
    {
      "version": "0.2.0",
      "tag": "@rushstack/rush-client-core_v0.2.0",
      "date": "Mon, 28 Sep 2026 20:08:26 GMT",
      "comments": {
        "none": [
          {
            "comment": "Publish the pre-bind test PID marker atomically so a starter cannot read an empty newly-created file as PID zero; retain single-daemon and ownership assertions."
          }
        ],
        "patch": [
          {
            "comment": "Add IConnectOrStartDaemonOptions.resolveStartCommandAsync so a start command is resolved only when a daemon must be started or replaced."
          },
          {
            "comment": "Reclaim provably stale daemon artifacts (a socket without an ownership record, a corrupt record, or a Linux PID reused since the record was written) instead of permanently disabling the daemon, and add resetDaemonArtifactsAsync() as the explicit recovery path referenced by fail-closed messages."
          },
          {
            "comment": "Retry daemon restarts with jittered backoff inside the admission deadline and return a `restartRetriesExhausted` fallback outcome instead of failing when successors keep restarting for other environments."
          },
          {
            "comment": "Return the daemon shutdown acknowledgement, including its optional active request count, from DaemonClient.shutdownAsync()."
          },
          {
            "comment": "Join detached startup helper closure without prematurely disconnecting IPC, keep its cwd outside the launched workspace, and drain failed launcher processes before reporting startup failure."
          },
          {
            "comment": "Reject in-process fallback after terminal control, even without a stdin source, while restoring raw mode and avoiding command replay."
          },
          {
            "comment": "Hand detached daemon startup to an independent helper with a durable pre-bind reservation. Prevent duplicate launches after the starting client dies, preserve explicit launch commands and ownership-based restart, and fail closed on ambiguous launcher failures."
          },
          {
            "comment": "Include the package MIT license and standard publication metadata, and enable the pack phase."
          },
          {
            "comment": "Reuse the validated native Windows LockFile exclusion for startup instead of maintaining a separate pipe mutex, preserving durable crash reservations and fail-closed handoff behavior."
          },
          {
            "comment": "Honor cancellation during daemon hello/pong readiness, close newly ready clients when cancellation wins, and preserve abort reasons instead of treating handshake failures as retryable."
          },
          {
            "comment": "Use an OS-owned Windows startup mutex and keep sharing-denied ownership reads unknown until the existing handoff deadline, preserving fail-closed startup and no-replay behavior."
          },
          {
            "comment": "Retry pre-request handshake resets within the startup deadline while a closing daemon retains resource ownership. Preserve fail-closed ownership checks and prohibit replay after execution begins."
          }
        ],
        "minor": [
          {
            "comment": "Support one ownership-attested pre-execution restart retry, cancellation during handoff, and native mutation protocol admission."
          },
          {
            "comment": "Introduce the opt-in daemon client with immutable requests, backpressured output, cancellation, raw input cleanup, private append-only launcher logs, serialized detached startup, and protocol-gated shutdown with bounded ownership-release coordination."
          },
          {
            "comment": "Capture the native invocation kind, reject Rushx peers older than protocol 0.8 before request submission, and prevent fallback after output or stdin admission."
          },
          {
            "comment": "Stream bounded stdin chunks only after admission and write credits, deliver ordered EOF, and preserve input for fallback to older peers."
          },
          {
            "comment": "Replace a mismatched daemon implementation under the startup mutex using an explicit launcher and verified ownership, before submitting any request."
          }
        ],
        "dependency": [
          {
            "comment": "Updating dependency \"@rushstack/node-core-library\" to `5.25.0`"
          },
          {
            "comment": "Updating dependency \"@rushstack/rush-daemon-protocol\" to `0.5.0`"
          },
          {
            "comment": "Updating dependency \"@rushstack/rush-daemon-transport\" to `0.4.0`"
          },
          {
            "comment": "Updating dependency \"@rushstack/heft\" to `1.3.2`"
          }
        ]
      }
    }
  ]
}