Figuring out performance bottlenecks


Maybe we could add another import modifier that treats the content of the import as if it was the import expression itself (like a C #include macro). The only restriction is that the content is a single import. Any thoughts ?


Perhaps this is its own discussion, but I think with thorough benchmarks it would be worth messing with the -A RTS flag.

See a comment on reddit here and the GHC user guide on gc options.


Interesting! Let’s better discuss this on the dhall-haskell issue tracker though.