I tried hacking on ghc years ago when it was still around 6.8. Everyone was helpful and friendly. But I just couldn't get any kind of dev env setup. Hopefully this work makes everything better for newbies, etc.
I still love haskell and call it my favorite language to this day, even though I barely use it as much as I had before. It's just to pleasing to use.
Yes the people are wonderful --- much nicer than the code! This is why it feels worth it to write something like this to try to steer such a big ship after so many years.
> Hopefully this work makes everything better for newbies, etc.
I'm partway through reading this. It is quite interesting. GHC has a standard disease of big old programs and refactoring the issues they describe will be a pretty impressive feat. I think doing something similar with GCC is likely hopeless.
I wonder if GHC has to be so complicated, to compile the language that it compiles, including all the extensions. JHC was a lot smaller but internally used a dependently typed intermediate language, that simplified some things. Of course it didn't implement nearly as complicated a language as GHC does.
(That was the originally submitted URL but in this case I think it's probably best if the paper is linked above.)