I was a little startled to learn that Nancy has its own razor implementation, that may or may not behave like razor. In practice, does this cause issues? What are "most people" using for a Nancy view engine? Why was the real razor not used?
First the easy answer. The Razor engine is, by far, the most downloaded view engine available for Nancy https://www.nuget.org/packages?q=nancy.viewengines
Now for the more complicated questions
Why was the real razor not used?
Because the "real" (and by real I'm going to assume that you mean the one used by the ASP.NET stack) Razor engine is tied to the HTTP abstractions that are built into the ASP.NET stack (HttpContext and all its friends) so there is no straight forward way to use it with Nancy.
The slightly longer answer for this is that you have to understand that Razor is really a parser and a Razor view engine is what sits in the middle of the consumer and the parser.
Nancy uses the Razor parser but we have to have our own view engine because that's what enabled Nancy to parse and execute Razor templates.
Now, it does get even more complicated. Many of the features you see in the ASP.NET Razor view engines, such as master pages, partials, various helpers, _ViewStart and so on, are not Razor (the parser) features, but they are an additional feature set that have been built into the view engine (you can almost think of it as middleware).
This means that for our engine we've had to re-implement much of these features because that's what's come to be expected from a Razor view engine.
I would like to point out that, if it was possible, then we would love to ditch our own implementation and use the one built by Microsoft (less code for us to maintain and it would mean we'd support 100% the same feature set), but unfortunately that's not our decision to make.. we can't take a dependency on their abstractions I am afraid
Hope this clears things up
/A