On Tue, 11 Aug 2026 03:18:50 GMT, Chen Liang <[email protected]> wrote:
> Given the complexities of handling buffering and state in `CharsetDecoder`, > should we consider an alternative where we support the known and most popular > stateless decoders, like ASCII, ISO8859-1, UTF8, and UTF16 without BOM? I think that my current approach pushed to Github today should tackle the complexity correctly now, so I would say explicitly supporting your mentioned favorites is a different story which might make sense to consider ontop in a follow-up JBS/PR. Having said that, I have not yet fully understood how it would *replace* dealing with the complexity, as still people could use custom charset decoders, which we cannot rule out. So I see addressing *specific* charset decoders as an *additional* speedup. Or did I missing your actual intention? 🤔 ------------- PR Comment: https://git.openjdk.org/jdk/pull/32264#issuecomment-5302476735
