@nanobpm/nano-coder 0.52.0 → 0.54.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (2) hide show
  1. package/README.md +19 -2
  2. package/package.json +5 -5
package/README.md CHANGED
@@ -758,8 +758,25 @@ content hash), so session logs stay small and `--resume` still works; a missing
758
758
  resume is sent as a text placeholder.
759
759
 
760
760
  Other per-provider fields: `replay_reasoning`, `max_tokens_param` (`max_tokens`, or `max_completion_tokens`
761
- which is the `openai` default), `retry_initial_backoff_ms`, `retry_max_backoff_ms` and
762
- `retryable_statuses`.
761
+ which is the `openai` default), `retry_initial_backoff_ms`, `retry_max_backoff_ms`,
762
+ `retryable_statuses` and `offline_budget_ms`.
763
+
764
+ A transport failure with no HTTP response (DNS, connection refused or reset, connect or
765
+ idle timeout) is treated as the network being down rather than a failed attempt: nano-coder
766
+ backs off (`retry_initial_backoff_ms` doubling to `retry_max_backoff_ms`) and keeps waiting,
767
+ reporting the outage on a single live line (`network unavailable (…); retrying in Ns
768
+ (Ctrl-C to stop)`) when stderr is a terminal, or announcing it once when output is piped, so
769
+ a long outage never floods the log. When a response finally arrives it announces the
770
+ recovery, so a turn survives a laptop sleep, a Wi-Fi change or a VPN reconnect instead of
771
+ failing after about a minute. A redirect-policy error (a redirect loop or a disallowed
772
+ redirect) is *not* an outage — HTTP is flowing — so it surfaces its actionable error at once
773
+ instead of being retried. These waits do not consume `max_retries` (which still bounds
774
+ HTTP-status retries like 429 or 5xx), and the Copilot session-token refresh waits the same
775
+ way. If a stream drops after it started replying, the partial reply is discarded (nothing
776
+ reaches history) and the request is re-sent, up to ten times per model call. The wait is
777
+ unbounded by default; set `offline_budget_ms` to give up after that many milliseconds
778
+ offline — the budget bounds the whole wait, backoff sleeps included, and no further request
779
+ is issued once it is spent.
763
780
 
764
781
  The old top-level `api_key` / `base_url` still work. They apply to `default_provider`,
765
782
  which becomes `openai` if it was `mock`.
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@nanobpm/nano-coder",
3
- "version": "0.52.0",
3
+ "version": "0.54.0",
4
4
  "description": "A 6MB coding agent. Run a fleet on your laptop.",
5
5
  "license": "Apache-2.0",
6
6
  "repository": {
@@ -27,9 +27,9 @@
27
27
  "node": ">=18"
28
28
  },
29
29
  "optionalDependencies": {
30
- "@nanobpm/nano-coder-darwin-arm64": "0.52.0",
31
- "@nanobpm/nano-coder-darwin-x64": "0.52.0",
32
- "@nanobpm/nano-coder-linux-arm64": "0.52.0",
33
- "@nanobpm/nano-coder-linux-x64": "0.52.0"
30
+ "@nanobpm/nano-coder-darwin-arm64": "0.54.0",
31
+ "@nanobpm/nano-coder-darwin-x64": "0.54.0",
32
+ "@nanobpm/nano-coder-linux-arm64": "0.54.0",
33
+ "@nanobpm/nano-coder-linux-x64": "0.54.0"
34
34
  }
35
35
  }