Sledge

> Rather than STT, safer alternatives would be to specialize this to one specific monad (was state the original motivation?) or to use a different algorithm that does not involve...

> `text` is a boot library, it cannot depend on `STMonadTrans` package, unfortunately. Removed. > I'm not very happy with the API of `Data.Text.Encoding`. The fact that the only way...

Ahhh... I see what you mean, and it'd be a problem with the list monad, too. However, I noticed that `streamDecodeUtf8With` returns a continuation in the `Decoding` data type. I...

While I understand the desire to use `Either`, that doesn't really get me closer to my goal, which is why I was going for arbitrary monad which would meet my...

Excellent! A couple of questions: * The byte position does not currently track across continuation invocations and instead resets back to zero. What's the opinion on whether it should track...

For the first question, I went ahead and let the library track the absolute position for the caller.

Stream decoders added for utf-16 and utf-32, and an `Either` decoder for ASCII. I'll work on the lazy options next.

This branch no longer uses simdutf anymore. Should all of it and references to it be removed, or is it worth modifying simdutf to have a `determine_utf8_prefix_length` function?

I suspect with the issue you pointed out of where the next to last `Word8` is invalid, this reimplementation will perform better, because the current implementation of `decodeUtf8With2` has the...

The error in the workflow seems to be due to an unrelated test. I've submitted an issue regarding it.