Part of #151. (Later than #162 per the 'architectural peers first' decision.)
Why. For an irrefutable cross-language comparison, include the servers that top TechEmpower plaintext/JSON.
Scope — new adapters (6-endpoint contract):
- uWebSockets.js (Node/C++) — JS TechEmpower leader.
- ASP.NET Core Kestrel (.NET minimal APIs) — TechEmpower top-5; currently a glaring omission.
- Vert.x or Netty (raw) (JVM) — JVM performance ceiling.
- drogon (C++) — top C++ contender.
- xitca-web (Rust) — current Rust top-tier (complements actix/axum/ntex/hyper).
Per adapter: own build dir + recipe, ready addr= handshake, contract endpoints, registry entry, feature flags. H1-first; H2/TLS where supported.
Depends on #164 (dotnet, JVM, node, cmake/clang for drogon).
Acceptance. Each is a column in static + concurrency (+ JSON) scenarios; methodology notes any framework-specific tuning (and that the same latitude was given to all).
Refs: servers/servers.go, servers/start*.go, cmd/runner/main.go (featureSetFor).
Part of #151. (Later than #162 per the 'architectural peers first' decision.)
Why. For an irrefutable cross-language comparison, include the servers that top TechEmpower plaintext/JSON.
Scope — new adapters (6-endpoint contract):
Per adapter: own build dir + recipe,
ready addr=handshake, contract endpoints, registry entry, feature flags. H1-first; H2/TLS where supported.Depends on #164 (dotnet, JVM, node, cmake/clang for drogon).
Acceptance. Each is a column in static + concurrency (+ JSON) scenarios; methodology notes any framework-specific tuning (and that the same latitude was given to all).
Refs:
servers/servers.go,servers/start*.go,cmd/runner/main.go(featureSetFor).