@declaw/sdk 1.2.2 → 1.2.3

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/CHANGELOG.md CHANGED
@@ -5,6 +5,16 @@ All notable changes to the Declaw TypeScript / JavaScript SDK are documented in
5
5
  The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
6
6
  and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
7
7
 
8
+ ## [1.2.3]
9
+
10
+ _Restore the conservative default connection ceiling._
11
+
12
+ ### Changed
13
+
14
+ - Reverted the default connection-pool ceiling back to 64 (from the 512
15
+ introduced in 1.2.2). High-concurrency workloads should opt in explicitly
16
+ via `DECLAW_SDK_CONNECTIONS`. No public API change.
17
+
8
18
  ## [1.2.2]
9
19
 
10
20
  _Higher default connection ceiling._
package/dist/index.cjs CHANGED
@@ -269,7 +269,7 @@ function getDispatcher() {
269
269
  }
270
270
  const undici = await import("undici");
271
271
  const connOverride = parseInt(process.env.DECLAW_SDK_CONNECTIONS || "", 10);
272
- const connections = Number.isFinite(connOverride) && connOverride > 0 ? connOverride : 512;
272
+ const connections = Number.isFinite(connOverride) && connOverride > 0 ? connOverride : 64;
273
273
  const streamsOverride = parseInt(process.env.DECLAW_SDK_MAX_CONCURRENT_STREAMS || "", 10);
274
274
  const maxConcurrentStreams = Number.isFinite(streamsOverride) && streamsOverride > 0 ? streamsOverride : 1e3;
275
275
  return new undici.Agent({
@@ -286,7 +286,9 @@ function getDispatcher() {
286
286
  // stream — silently defeating the H2 multiplexing that allowH2 enables.
287
287
  // Omitted, H2 sessions default to unlimited concurrent streams (capped
288
288
  // by maxConcurrentStreams) and H1.1 fallback defaults to 1 (safe, no
289
- // head-of-line risk on non-idempotent POSTs).
289
+ // head-of-line risk on non-idempotent POSTs). This is the single
290
+ // biggest burst-latency lever: with pipelining:1 a 100-concurrent burst
291
+ // queued ~36 requests behind the `connections` cap.
290
292
  allowH2: true,
291
293
  connect: { keepAlive: true, keepAliveInitialDelay: 5e3 }
292
294
  });