It could be a feature not a bug
Heading through the fourth decade and the default Haskell type for representing text is still the venerable, naive
[Char] with it’s dire continental-drift performance. Instead of sorting this all out we have at least five different central types for representing text.
That last idea is a beautiful, brilliant proposal though:
newtype-wrapping a Utf8-encoded
ByteString builder. Everything is right there hiding in plain sight, just requiring some gentle organisation.
Am I buying it? Not quite.
ByteString is great for bulk reading, writing and processing of text, which is almost certainly UTF8-encoded. But a proper array of Unicode characters, as provided by text, is a cleaner abstraction and more preferable for all those everyday situations that are not I/O dominated.
We just don’t seem to be able get up the escape velocity. Maybe we are doomed to inhabit this circle of hell for eternity. Better get comfortable then and turn the bug into a feature — not the first time Haskellers would have done that!
Got an issue with any of this? Please share or drop me a line (see below).