[Ur] Combinator Parsing in Ur: Anonymous function remains at code generation

[Adam Chlipala]

On 07/23/2014 06:19 PM, orchidaceae phalaenopsis wrote:
> It is unintuitive that Ur/Web doesn't currently accept code when it
> cannot optimize out all anonymous functions. I hope that there will be
> some way to fully support higher order functions in future, would you
> be able to comment on this please?

I expect a future Ur version to support first-class functions at 
run-time in server-side code.  We're working on meta-level tools to make 
it more pleasant to build the next clean-slate Ur reimplementation, but 
there is no concrete schedule.

> It would also be nice if it was
> pointed out more loudly that this is the case in manuals and such.
>    

I've added a forward reference to this effect, at the end of the manual 
section on invoking the compiler.

> Of course if you can do first-order programming you can implement an
> interpreter for a higher order language, so I got the idea of
> embedding a language that lets one write parsers into Ur/Web:
>    

I hope such an extreme solution isn't necessary. :)

I don't currently see any obstacle to using standard parser combinators 
with Ur/Web; you've just been running into compiler bugs, by virtue of 
venturing into unexplored territory.