[Haskell-cafe] Re: Compile-time here document facility

[Dave Bayer]

Donald Bruce Stewart <dons <at> cse.unsw.edu.au> writes:

> bayer:
> > I couldn't find a compile-time here document facility, so I wrote one  
> > using Template Haskell:
> 
> Very nice! You should wrap it in a little .cabal file, and upload it to
> hackage.haskell.org, so we don't forget about it.
> 
> Details on cabalising and uploading here:
> 
>     http://haskell.org/haskellwiki/How_to_write_a_Haskell_program
>     http://cgi.cse.unsw.edu.au/~dons/blog/2006/12/11

I was waiting for the other shoe to drop, as usual. I tweaked the code a
bit, and I'm getting ready to contribute it to Hackage, as I find it
useful. The current source can be found at

http://www.math.columbia.edu/~bayer/Haskell/Annote/HereDocs.html

Basically, it bothered me that I was using assert to figure out the file
name (an idea that others have also had) because asserts disappear under
optimization. I looked again at Control.Exception, and I realized that I
would be better off catching a pattern-matching exception:

    $(hereDocs $ let [e] = [] in e)

(Can anyone come up with a shorter pattern match failure, that can be used
as a value here?)

Of course, anything like this is like gas fireplace logs, pointlessly
recreating the past. There is little wrong with simply hard-coding the file
name; GHC has this nice way of letting me know when code like this gets out
of sync:

    $(hereDocs "Strings.hs")

So here is my question:

> The source should live under a directory path that fits into the existing
> module layout guide.
> http://www.haskell.org/~simonmar/lib-hierarchy.html

A here document is a kind of data, but it is really a language extension,
and one that depends on a GHC extension, Template Haskell. I'd go for

    Data.HereDocs

but that doesn't seem quite right. The newbie in me doesn't want to park my
car in your living room, so to speak. What's the right place for this?