Niklas Saari

I see. What would you think if I try to do it as: ```rust pub enum Integer { Primitive(P), Variable(Box), } ``` That would make it easier to use internally,...

> > That would make it easier to use internally, e.g. if we use primitive types, like u64 when defining ASN.1 types, then for optimal usage they would need to...

Maybe the title was was bit misleading in that sense. This PR attempted to solve all cases of using integers, particularly the issue with constraints which are defined by using...

I would have appreciated a bit more information, as it also currently uses the trait in this PR as your comment suggests, but just for variants of enum instead of...

I think it is quite clear now. There was misunderstanding with the constrained and variable-sized integers. I though that with `isize` we try to optimize both cases. This PR needs...

There is one, maybe minor behavior left which should be decided (just about how `PartialEq`/`Eq` works, or what is the level of performance optimization we should consider) for ```rust #[derive(Debug,...

I believe this can be reviewed at this point. Everything integer related is now moved into `types/integer.rs`. Closes #308 #296 Some initial bench results below based on [this](https://github.com/librasn/rasn/blob/0ff9f8ba1e71f94da2b979ba27a8f9ac6c4ce033/benches/integer.rs) . UPER...

I have a working branch which introduces new lifetime type for `Encoder` and allows reusing the same buffer recursively quite much with the new `AnyEncoder` associated type, without `Rc` or...

This might a bit unfair comparision since Vec might be as fast as it gets, if you don't need some features of `Bytes` type. `Bytes` type has an advantage that...

Just a note, OER has now `encode_buf` for `&mut Vec` type for starters. There is internally second buffer for length calculations but other than that, allocations are now almost at...