For me, the biggest downside is lack of solid embedded device support -- arduino (Atmel AVR), android(ARM), iOS (ARM). After using Haskell pretty much full-tim

For me, the biggest downside is lack of solid embedded device support -- arduino (Atmel AVR), android(ARM), iOS (ARM). After using Haskell pretty much full-time for 10 years, writing C and Java code makes me sad. The support for the above mentioned platforms is in-progress, but is not yet mature. There are some neat things like Atom which use a Haskell DSL to target arduino. My other issue is that the garbage collector in GHC is not really sufficient for real-time audio applications because it can pause for too long. GHC HQ has tried to tackle this in the past -- but there is a reason why it is a research topic :) If your application requires interfacing to a C++ world -- your are not going to have fun. Though there might be a GSoC project for that this summer? Also, GUI stuff is somewhat lackluster. There are bindings to gtk, etc. And they can get the job done. But they don't really capture the essence of what makes Haskell awesome. We are still searching for the GUI abstraction that really clicks.