Writing kernel modules in Haskell

Posted Sep 17, 2009 2:51 UTC (Thu) by Pseudonym (guest, #60861)
In reply to: Writing kernel modules in Haskell by jordanb
Parent article: Writing kernel modules in Haskell

Of course. The issue here is essentially how you want to divide up work between the programmer and the implementation.

There's always the risk of your GC not having enough time to reclaim all of memory, but that is a risk with a non-GC'd implementation, too.

First off, deallocating memory is always more expensive than allocating it in a concurrent environment. You can allocate memory from whichever region has the lowest contention, but you can only deallocate it to the place that it came from.

Secondly, if your time constraints were that tight, then you probably didn't have enough time to manually free all that memory right now, either.

Yes, there's always the possibility of your program misbehaving. But in the real-time space, that's always a risk. Using a GC'd implementation doesn't free you from hard thinking, it just frees you from bothering with a certain amount of boilerplate.