Release 2.8.0-preview.2

[mtmk]

Preview release bumping version to 2.8.0-preview.2. Adds nats-server v2.14 support for fast-ingest batch publishing via the new AllowBatchPublish field on StreamConfig (ADR-50). Also picks up Pin ID handling fix for duplicate status headers, README rewrite, and CI/test flake fixes.

• Add AllowBatchPublish stream config field (Add AllowBatchPublish stream config field #1120)
• Bump OpenTelemetry and OpenTelemetry.Exporter.OpenTelemetryProtocol (Bump OpenTelemetry and OpenTelemetry.Exporter.OpenTelemetryProtocol #1121)
• Fix setting wrong pin id from status header (Make pinned client Pin ID handling tolerant of duplicate headers #1116)
• Fix slow-consumer test first-ping RTT flap (Fix slow-consumer test first-ping RTT flap #1115)
• Fix net481 TLS test flakes (Fix net481 TLS test flakes #1111)
• Rewrite README intro (Rewrite README intro #1114)

[synadia-claude-reviewer]

Claude finished @mtmk's task in 3m 7s —— View job

---

PR Review: Release 2.8.0-preview.2

• Gather context
• Review PR diff and version change
• Check associated commits/changes
• Post review summary

---

This is a release bump PR (version.txt: 2.8.0-preview.1 → 2.8.0-preview.2). The one-line diff is trivially correct. I reviewed the key changes included in this release that are already in the branch.

AllowBatchPublish (StreamConfig.cs:309)

The field is well-formed:

[JsonPropertyName("allow_batched")]
[JsonIgnore(Condition = JsonIgnoreCondition.WhenWritingDefault)]
public bool AllowBatchPublish { get; set; }

WhenWritingDefault is correct for a bool — false won't be serialized, matching the server's field-absence semantics. The test at ManageStreamTest.cs:300 correctly carries [SkipIfNatsServer(versionEarlierThan: "2.14")].

Minor observation: The JSON property name allow_batched is slightly inconsistent with the existing AllowAtomicPublish → allow_atomic pattern (one uses an adjective, the other uses a past participle). This is presumably dictated by the server-side ADR-50 field name and not something the client controls, but worth a quick cross-check against the server spec if it hasn't been done already.

Pin ID fix (NatsJSExtensionsInternal.cs, NatsJSConsume.cs:459, NatsJSFetch.cs:292)

TrySetPinIdFromHeaders is correctly invoked only in the regular-message branch (the else path after all status-header checks), preventing spurious pin ID overwrite from idle-heartbeat or other status responses. The lock-based GetPinId/SetPinId in NatsJSConsumer.cs:326-343 is correct for the read-heavy, infrequent-write pattern here.

TryGetLastValue uses values[^1] which is fine — in practice a Nats-Pin-Id header will have exactly one value, but taking the last is the safe choice for multi-value headers.

Version bump

2.8.0-preview.1 → 2.8.0-preview.2 is correct and matches the PR title and description.

Overall: No correctness or security issues found. The release looks good to merge once CI passes.

[MauriceVanVeen]

LGTM