Document backend capacity management

This commit is contained in:
2026-07-29 21:16:49 +00:00
parent d2c4051dd0
commit e61ab700c7
10 changed files with 269 additions and 57 deletions

View File

@@ -1,6 +1,6 @@
# Backend-Specific Concurrency Management
**Status:** Accepted.
**Status:** Complete.
## Purpose

View File

@@ -1,6 +1,6 @@
# Backend-Specific Concurrency Management Implementation Plan
**Status:** Ready for implementation.
**Status:** Complete.
## Purpose
@@ -647,7 +647,7 @@ releases admission, and existing preparation semantics remain unchanged.
## Stage 4 — Engine Assembly And Public Runtime Contract
**Status:** Pending.
**Status:** Complete.
### Goal
@@ -721,7 +721,7 @@ through the same wrapper.
## Stage 5 — Durable Documentation And Final Validation
**Status:** Pending.
**Status:** Complete.
### Goal
@@ -819,6 +819,19 @@ limits and cancellation safety, durable contracts no longer depend on roadmap
prose, and the OpenRouter compatibility change is clearly reported for the
next minor release.
## Implementation Handoff
Backend-specific capacity management is implemented and has passed the complete
repository validation sequence. The built-in OpenRouter backend now permits 16
active generations and a waiting capacity of 1024. Custom backends remain
unlimited when their limit is omitted, and endpoint-only profiles remain
unlimited.
Publishing this behavior requires a pre-`v1` minor release. Its release notes
must identify that unusually high concurrent OpenRouter use can now wait or
return `ErrCapacityExceeded`. This implementation does not change a module
version or create a tag.
## Open Questions
None. The feature roadmap and this plan fix the public representation,