[efficiency-improver] perf: encode JSON body once in TcpMessageHandler.WriteRequestAsync

[Evangelink]

🤖 This PR was created by Efficiency Improver, an automated AI assistant focused on reducing energy consumption and computational footprint.

Goal and Rationale

In TcpMessageHandler.WriteRequestAsync, the serialized JSON message string was processed in two separate steps:

1. Encoding.UTF8.GetByteCount(messageStr) — scans the string to compute the byte count for the Content-Length header
2. StreamWriter.WriteAsync(messageStr) — internally re-encodes the string to UTF-8 bytes for writing

This PR replaces the two-step approach: on netstandard2.0 we call Encoding.UTF8.GetBytes(messageStr) once (getting both bytes and length), and on NETCOREAPP we call GetByteCount + GetBytes into a pooled ArrayPool<byte> buffer, then write directly to the underlying stream in both cases.

Focus Area

Code-Level Efficiency — redundant CPU work (double string encoding) and unnecessary StreamWriter buffering in the server-mode JSON-RPC message write path.

Approach

NETCOREAPP path (before → after):

• Before: GetByteCount(messageStr) (1 scan) + StreamWriter.WriteAsync(messageStr) (1 internal encode + buffer alloc) = 2 scans, StreamWriter heap buffer
• After: GetByteCount + GetBytes into ArrayPool<byte> buffer (2 scans, 0 heap allocs) + direct BaseStream.WriteAsync = same 2 scans, 0 heap allocations

netstandard2.0 path (before → after):

• Before: GetByteCount(messageStr) (1 scan) + StreamWriter.WriteAsync(messageStr) (1 internal encode + alloc) = 2 scans, 2 allocations
• After: Encoding.UTF8.GetBytes(messageStr) (1 scan, 1 byte[] allocation) + direct BaseStream.WriteAsync = 1 scan, 1 allocation

The key improvement on netstandard2.0 is eliminating one full string scan. On NETCOREAPP, we eliminate the StreamWriter's internal char→byte transcoding buffer heap allocation.

Energy Efficiency Evidence

Proxy metric: CPU instruction count (fewer string traversals) and heap allocation (fewer GC-triggering byte buffer allocations = less DRAM energy from GC pressure).

Call frequency: WriteRequestAsync is called once per JSON-RPC message in server mode (IDE-driven test runs via --server). Every test result, discovery result, and progress heartbeat goes through this path. In a large test suite, this fires thousands of times per session.

Green Software Foundation — Hardware Efficiency: removing a full string re-traversal per RPC message reduces CPU work per functional unit (one message delivered). Fewer heap allocations reduce GC-induced stop-the-world pauses that waste proportional CPU energy.

Trade-offs

The code is slightly more complex due to the ArrayPool pattern on NETCOREAPP. The inline comment explains the rationale. The wire format is identical — the same UTF-8-encoded JSON body is written to the TCP stream.

Test Status

✅ Build succeeded
✅ All unit tests passed (./build.sh -test)

Generated by Efficiency Improver · sonnet46 3.6M · ◷

Add this agentic workflows to your repo

To install this agentic workflow, run

gh aw add githubnext/agentics/workflows/efficiency-improver.md@main

[Copilot]

Pull request overview

This PR optimizes the server-mode JSON-RPC write path by encoding serialized request bodies to UTF-8 bytes explicitly before writing them to the TCP stream.

Changes:

• Replaces StreamWriter.WriteAsync(messageStr) for the JSON body with direct writes to the underlying stream.
• Uses ArrayPool<byte> on NETCOREAPP targets and a single GetBytes allocation on non-NETCOREAPP targets.
• Preserves existing header formatting and flushing behavior while reducing body-write overhead.

Show a summary per file

File	Description
src/Platform/Microsoft.Testing.Platform/ServerMode/JsonRpc/TcpMessageHandler.cs	Updates WriteRequestAsync to pre-encode the JSON body and write bytes directly to the TCP stream.

Copilot's findings

• Files reviewed: 1/1 changed files
• Comments generated: 1