Problem
The shared codec always writes UTF-8 JSON, but the Redis client decodes replies with the configured encoding. With encoding="latin-1", cached bytes are corrupted on the way back (b"\xe9" returns as b"\xc3\xa9").
Proposal
- 0.3.8: warn (or reject) when
encoding is not UTF-8, and document it in BACKENDS.md.
- 0.4.0: drop the parameter and read raw bytes (tracked separately).
Problem
The shared codec always writes UTF-8 JSON, but the Redis client decodes replies with the configured
encoding. Withencoding="latin-1", cached bytes are corrupted on the way back (b"\xe9"returns asb"\xc3\xa9").Proposal
encodingis not UTF-8, and document it in BACKENDS.md.