mirror of
https://github.com/PyMySQL/mysqlclient.git
synced 2025-08-15 11:10:58 +08:00
3fb4eb1b11
Before *Most* string fields were passed to custom converter functions as "bytes", everything else was assumed to be valid encoded utf-8 and passed as str. This could cause issues where some unknown field was not valid utf-8 and a decode error is triggered where it can't be fixed or avoided by the end user. Instead we should short list those fields that we know for sure have to be str type because the existing converters expect it to be that way. So all the date conversions, and (NEW)DECIMAL since the python decimal type will not accept bytes object. Everything else stays bytes, which is the least suprising and safest thing to do.