Accept non-Haskell QuasiQuotes

[ivan-m]

Rather than throwing an error (thus making it impossible to use shm in a
function that has a QQ somewhere in it), instead just gracefully fail.

Ideally, the contents of the QuasiQuote would be one context contained
within the overall QQ, but I couldn't figure out how to do that.

Closes #116.

[ivan-m]

I think the build error is because the version of GHC on Travis-CI is too old.

[ivan-m]

Any chance of getting this merged in? I'm dealing with a codebase that is rather liberal with its usage of quasiquotation for SQL and general string interpolation, and without this patch it makes it very difficult to hack upon.

[chrisdone]

Can you give me an example so that I can test this out? Generally SHM works well with quasiquotes.

[ivan-m]

I gave a sample one in #116. Not specified in that one, but the interpolation syntax we use is $n for show n, $(foo n) for show (foo n) and ${bar n} for bar n (which evaluates to a String); I think it was the last one that caused problems.

(I can try and get a better example at work tomorrow if this doesn't provide enough of an example.)

[chrisdone]

I'm testing this out now.

[chrisdone]

I seem to come back to this from time to time and forget why it's not merged. Last I tried it, it introduced some other parser issue I think.