On 06/10/2011 16:34, Chris Smith wrote:
Simon, thank you! That makes sense then.
I'd missed the fact that including the entire top-level scope requires the module to be interpreted. I suppose the "right" thing to do would be to not do that; but sadly, that seems to also mean that modules without a 'module Foo where' only export the single symbol 'main', and I liked the idea of students in my class not having to write out the module bit explicitly.
So I've been actually using interpreted code the whole time? If so, I suppose there's no loss in taking out HscAsm entirely, then!
Actually you haven't been using interpreted code, but you haven't been getting the full top-level scope of the compiled modules. I just tried a little test with ghci -fobject-code, and found that something is definitely amiss with this combination, so I created a ticket: http://hackage.haskell.org/trac/ghc/ticket/5534 Cheers, Simon