Hi,

On top of being more symmetrical with decode_from_slice, this is very handy for cases where that information is not easily available decoding side.

Use case that I'm having is where the slice host a collection of bincode-encoded items. The overall size is know (through a specific header) but not individual item encoded sizes. A representative example could be a list of &str.

That’s obviously fixable through a per item extra usize header but make encoding/decoding slower, and serialized size larger, than need to be.

I might have missed something but it seem decoder's reader is having that information which could gain from being exposed (as for
decode_from_slice).

And thanks a lot for the very nice crate!

Seb