[Ur] Feedback and questions

Adam Chlipala adamc at impredicative.com
Thu Dec 24 08:22:35 EST 2009


Artyom Shalkhakov wrote:
>> I'm not planning that for any time soon. Since the only compiler is
>> whole-program, a REPL would require significant extra implementation.
>>      
> Hm, okay. Is it possible to invoke the compiler for syntax analysis
> and type-checking then? Something like "urweb --typecheck foo".
>    

You can already get that with the Ur/Web library loaded in SML/NJ, via:
     Compiler.check Compiler.toElaborate "foo";
I'll add a way to do that with the command-line tool, too.

> What would you say about "Type theory and
> functional programming" by Simon Thompson?
>    

I've never read it, so I don't have anything to say about it.

> Speaking of error messages, sometimes they are extraordinarily long.
> OTOH, knowing just the error location is enough most of the time.
>    

Well, for now, I'm erring on the side of including enough information.  
Concrete suggestions of algorithms for improving error message 
generation are welcome.



More information about the Ur mailing list