This may have gone undetected in the past because people had latin1 databases, but we do a lot to push people toward UTF-8 which makes index size 3 bytes per "char". The answer is either to reduce it to String(255) or change it to a VARBINARY. The former is preferred, as 256 byte varchars become TINYTEXT which is less efficient as it is stored basically as a separate row.
This may have gone undetected in the past because people had latin1 databases, but we do a lot to push people toward UTF-8 which makes index size 3 bytes per "char". The answer is either to reduce it to String(255) or change it to a VARBINARY. The former is preferred, as 256 byte varchars become TINYTEXT which is less efficient as it is stored basically as a separate row.