One of the things I've wanted to do for the 3.0 release is embrace Span<T> in IPattern<T>. I've got a prototype to look at how we could do this with another interface in a 3.1 if we wanted to, but it would be nicer to have it in IPattern<T> to start with.
However, that has the distinct disadvantage that it's blocking the 3.0 release, and there are features (e.g. YearMonth and the TypeConverters) that aren't in 2.x at the moment. One option would be to look through all the new features and bug fixes that we've introduced in the master branch since splitting off 2.4, and then deciding which to apply for a 2.5 release. Obviously that's a significant chunk of work in itself, and it may be better just to get my head down into the Span<T> work to optimize it further, so we could release 3.0. (The main blocker at the moment is that the parsing is slower with a string, which is annoying and surprising.)
So to summarize, there are three options:
- Release 3.0 soon without Span<T> support, introducing that in a new interface in 3.1
- Release 2.5 with new features, so that there's less urgency for 3.0
- Focus on getting Span<T> support into 3.0 and release it
Any thoughts on any of these options?