On Mon, Oct 25, 2004 at 02:14:28PM +0100, Simon Marlow wrote:
On 24 October 2004 20:51, Sven Panne wrote:
IMHO it would be best to use explicit bracketing where possible, and hope for the RTS/GC to try its best when one runs out of a given resource. Admittedly the current Haskell implementations could be improved a little bit in the last respect.
Indeed, GHC could/should try to free up file descriptors when they run out. It's a bit tricky though. Hm, I'm not sure about the "should". Garbage collection is meant for memory, and anything making that less clear makes people more likely to depend on incorrect assumptions. And redefining GC to be a collection of _all_ garbage, instead of just memory doesn't sound so fantastic either.
At the moment performGC doesn't actually run any finalizers. It schedules a thread to run the finalizers, and you hope the thread runs soon. So if you're running performGC for the purposes of finalization, then almost certainly (performGC >> yield) is better. I've been wondering whether having a more synchronous kind of finalizer would be a good thing. Synchronous finalizers seem to be difficult to implement in e.g. the JVM, so may make a JVM-backend more difficult. (I'm thinking about how CPython vs Jython go about closing file-handles... CPython uses (primarily) reference-counting, so files get closed as soon as they aren't referenced anymore, which lots of people now depend on. Jython uses Java-GC, so some CPython programs may suddenly fail...)
Nevertheless, I agree with the general sentiment on this thread that file descriptors shouldn't be treated as a resource in the same way as memory.
Cheers, Simon
Groeten, Remi P.S. Why do so many people (including me) seem to come to Haskell from Python? It can't be just the indentation, can it? ;) -- Nobody can be exactly like me. Even I have trouble doing it.