Custom UTF-8 decoder caused 32x slowdown and OOM crashes in Node.js service

A developer profiling a Node.js service discovered that processing three concurrent 18.7 MB JSON-RPC responses crashed the process due to out-of-memory errors on a 1 GB heap. The culprit was a custom UTF-8 decoder that created nearly 20 million intermediate array elements and individual character strings per response, consuming up to 794 MB of heap each. Replacing the custom function with the native TextDecoder API reduced processing time from 928ms to 10ms and heap usage from 794 MB to 47.6 MB per response. The fix was a single line of code, with the fatal: true option preserving the original error-handling behavior for invalid UTF-8 input. The case highlights how technically correct custom implementations of platform-native operations can become severe performance liabilities at scale.