You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Is it possible to avoid using a pipe for communicating between C to Haskell? A pipe uses 2 fds. It will be much more scalable if we can do away with those fds.
Haskell stack runs out of fds on OSX while using filewatch via fsnotify. Please see commercialhaskell/stack#1979 .
The text was updated successfully, but these errors were encountered:
hmm, perhaps I can change it to use a single pipe for all watches. if many watches are created, having an thread for each of them would probably also be less than ideal.
Haskell threads should be relatively cheap but pthreads on the C side might not scale well. Can we make watchCallback a Haskell function and use the C wrapper of the Haskell function in the C code? GHC supports that. That way we can avoid the pipes as well as pthreads, making the library scalable irrespective of how the application designers use it.
Is it possible to avoid using a pipe for communicating between C to Haskell? A pipe uses 2 fds. It will be much more scalable if we can do away with those fds.
Haskell stack runs out of fds on OSX while using filewatch via
fsnotify
. Please see commercialhaskell/stack#1979 .The text was updated successfully, but these errors were encountered: