On Fri, Dec 13, 2013 at 06:29:34AM +0100, Wladimir wrote:
> ...
> I agree that it would be nice if the consensus rules were better
> documented, or if the appropriate consensus-critical parts were isolated
> from the reference client and converted into a library. So why doesn't a
> project to that? Why this focus on re-implementing before
> documenting/understanding?
> ...

It is a common misunderstanding that isolating consensus-critical parts
into a separate library woud do any good (except for the niceties of
logical separation). If the entire backbone used this library, then the
entire backbone would still rely on the same code.

It is true that having a rigid specification could help things. However,
as long as there is only one implementation of it, there is little
incentive to have a specification at all. I'm talking philosophy here,
there have been >1 implementations by now.

Re-implementing is a way of understanding.