instar 1.3.405 → 1.3.406
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.
- package/dist/commands/server.d.ts.map +1 -1
- package/dist/commands/server.js +25 -8
- package/dist/commands/server.js.map +1 -1
- package/package.json +1 -1
- package/src/data/builtin-manifest.json +2 -2
- package/upgrades/1.3.406.md +25 -0
- package/upgrades/side-effects/pool-stream-connector-observability.md +53 -0
package/package.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
|
1
1
|
{
|
|
2
2
|
"$schema": "./builtin-manifest.schema.json",
|
|
3
3
|
"schemaVersion": 1,
|
|
4
|
-
"generatedAt": "2026-06-07T17:
|
|
5
|
-
"instarVersion": "1.3.
|
|
4
|
+
"generatedAt": "2026-06-07T17:20:53.176Z",
|
|
5
|
+
"instarVersion": "1.3.406",
|
|
6
6
|
"entryCount": 199,
|
|
7
7
|
"entries": {
|
|
8
8
|
"hook:session-start": {
|
|
@@ -0,0 +1,25 @@
|
|
|
1
|
+
# Upgrade Guide — vNEXT
|
|
2
|
+
|
|
3
|
+
<!-- assembled-by: assemble-next-md -->
|
|
4
|
+
<!-- bump: patch -->
|
|
5
|
+
|
|
6
|
+
## What Changed
|
|
7
|
+
|
|
8
|
+
Live verification of cross-machine dashboard streaming found the request starting but no terminal output ever arriving — and no error either: the connector that opens the upstream link to a peer swallowed every outcome silently, and its ticket-mint call had no explicit timeout, so a wedged mint could hang indefinitely (the stream never opened AND never errored, leaving the dashboard silent forever). This adds (a) an explicit 10s timeout on the mint so it fails honestly instead of hanging — a bounded failure surfaces peer-stream-lost / unreachable to the user — and (b) per-step logging (mint attempt/result, upstream open/close/error) on both the requesting connector and the serving mint verb, so the live path is finally observable.
|
|
9
|
+
|
|
10
|
+
## What to Tell Your User
|
|
11
|
+
|
|
12
|
+
Nothing user-facing — this makes the cross-machine streaming path debuggable and stops a silent hang. (The click-to-stream feature itself is being finished; this is the diagnostic that unblocks it.)
|
|
13
|
+
|
|
14
|
+
- audience: agent-only
|
|
15
|
+
- maturity: stable
|
|
16
|
+
|
|
17
|
+
## Summary of New Capabilities
|
|
18
|
+
|
|
19
|
+
- Explicit 10s timeout on the connector's `pool-stream-ticket` mint (no more silent indefinite hang).
|
|
20
|
+
- Per-step connector logging (`[pool-stream-connector]`) + serving-side mint-verb logging (`[pool-stream-ticket]`).
|
|
21
|
+
|
|
22
|
+
## Evidence
|
|
23
|
+
|
|
24
|
+
- Live-verify (2026-06-07): laptop /ws remote subscribe returned `subscribed` then 22s of silence (no output, no error); the Mini had no `stream-tickets.json` (never minted). Root: silent-hang + no observability. This fix makes the failure visible and bounded; re-verify reads the new logs.
|
|
25
|
+
- tsc clean. Behavior change is limited to bounding a previously-unbounded call + logging.
|
|
@@ -0,0 +1,53 @@
|
|
|
1
|
+
# Side-Effects Review — Pool-stream connector observability + bounded mint
|
|
2
|
+
|
|
3
|
+
**Version / slug:** `pool-stream-connector-observability`
|
|
4
|
+
**Date:** `2026-06-07`
|
|
5
|
+
**Author:** `echo`
|
|
6
|
+
**Second-pass reviewer:** `not required`
|
|
7
|
+
|
|
8
|
+
## Summary of the change
|
|
9
|
+
|
|
10
|
+
Adds an explicit 10s timeout to the requesting connector's pool-stream-ticket
|
|
11
|
+
mint (was unbounded → could hang silently) and per-step logging to both the
|
|
12
|
+
connector and the serving mint verb. Diagnostic + robustness for the live
|
|
13
|
+
cross-machine streaming path; found by live-verify (no output, no error,
|
|
14
|
+
no ticket minted on the peer).
|
|
15
|
+
|
|
16
|
+
## Decision-point inventory
|
|
17
|
+
|
|
18
|
+
One: bound the mint with a timeout (10s) so a wedge fails honestly →
|
|
19
|
+
peer-stream-lost/unreachable, instead of hanging the stream forever silently.
|
|
20
|
+
|
|
21
|
+
## 1. Over-block
|
|
22
|
+
|
|
23
|
+
A genuinely-slow mint (>10s) now fails where it might previously have eventually
|
|
24
|
+
succeeded — acceptable: 10s is far beyond a healthy mesh round-trip, and an
|
|
25
|
+
honest failure the user can retry beats an indefinite silent hang.
|
|
26
|
+
|
|
27
|
+
## 2. Under-block
|
|
28
|
+
|
|
29
|
+
Logging only observes; the timeout only bounds. No new gating.
|
|
30
|
+
|
|
31
|
+
## 3. Level-of-abstraction fit
|
|
32
|
+
|
|
33
|
+
The timeout uses the existing MeshRpcClient per-call `{ timeoutMs }` (the T1
|
|
34
|
+
mechanism). Logging matches the `[subsystem]` dim-log idiom used across the
|
|
35
|
+
mesh wiring.
|
|
36
|
+
|
|
37
|
+
## 4. Signal vs authority compliance
|
|
38
|
+
|
|
39
|
+
**Required reference:** [docs/signal-vs-authority.md](../../docs/signal-vs-authority.md)
|
|
40
|
+
|
|
41
|
+
This REMOVES a silent-swallow (the connector previously hid every failure) —
|
|
42
|
+
moving toward the no-silent-fallbacks standard, not away.
|
|
43
|
+
|
|
44
|
+
## 5. Interactions
|
|
45
|
+
|
|
46
|
+
- PeerStreamProxy: a bounded mint failure now reliably reaches onClose →
|
|
47
|
+
peer-stream-lost → bounded reconnect → machine-unreachable (the designed
|
|
48
|
+
path, previously unreachable when the mint hung).
|
|
49
|
+
- No change to the serving endpoint, ticket store, or WSManager routing.
|
|
50
|
+
|
|
51
|
+
## 6. External surfaces
|
|
52
|
+
|
|
53
|
+
Log lines only. No route/config/notification change.
|