Error with '#' at beginning of line in literate Haskell

[guest]

When trying to load the attached file, I'm getting this error:

lexical error at character '\n'

pointing to the position of the first '#' in the file.

[simonmar]

GHC's unlit preprocessor leaves lines beginning with '#' in the output, so that you can use CPP directives in literate comments. It has always done that.

Arguably CPP should be run before unlit. I don't think I have a strong opinion, but it's entirely possible that would lead to some unforseen consequences; someone would need to try it.

[guest]

Replying to simonmar:

I (the bug reporter) have no strong opinion either, but at least the error message could perhaps be made more helpful? I had no idea what the problem was (although it is obvious now that I know the reason), so I googled somewhat and found nothing, then I asked at Freenode's #haskell channel and John Millikin suggested I file a GHC bug, so here it is... Thanks for the feedback.

[simonmar]

#4836 is a dup of this, but has more useful info so let's keep that one instead.