[Haskell-cafe] Template Haskell a Permanent solution?

[Jonathan Geddes]

On Mon, Dec 27, 2010 at 12:44 AM, Henning Thielemann
<lemming at henning-thielemann.de> wrote:
> I think it would be enough, if the compiler could be told to unfold an
> expression like
>  parse "text in a domain specific language"
>  at compile time.

I'm afraid I have to disagree with you here. Being able to specify
that the string should be parsed at compile time is only half of the
equation in my mind. The other half is the clean syntax for multi-line
strings.

Haskell already has great syntax for specifying data in a declarative
manner. Especially in contrast with ie Java/C++. Even as good as the
dynamic languages ie JavaScript/Python/Ruby. When you add the ability
to specify data in ANY syntax you can parse, Haskell is clearly the
best. But the complexity of TH detracts from the elegance of this
greatly in my opinion. And wrapping your "data" in string syntax,
multi-line or otherwise, detracts from the elegance as well. A syntax
has to be less painful or more convenient or more
readable/maintainable than literal list/record syntax before it is
useful.