Haskell Platform decision: time to bless parsec 3?

Philippa Cowderoy flippa at flippac.org
Sat Nov 6 14:18:39 EDT 2010

On 06/11/2010 17:47, Stephen Tetley wrote:
> Couldn't an independent package could achieve this goal?
> If the combinators were defined in Parsec they would be under the
> Parsec namespace rather than Control.* and Parsec having type
> specialized versions might produce better error messages.

Not just error messages, either - it's room for further optimisation 
possibilities any time a combinator might expand to an infinite grammar, 
for example. And Parsec already tells you at runtime when it has certain 
kinds of bad grammar that it can't spot statically.

flippa at flippac.org

More information about the Libraries mailing list