Simon Marlow wrote:
The test program works after applying the following patch to GHC that disables nonblocking mode after the fork and before the exec.
If we did this, we would also need to re-enable non-blocking mode after the rawSystem completes. Also, this doesn't work well if runProcess is used and the Haskell program continues to run while the subprocess is executing, because the O_NONBLOCK flag is shared between parent and child.
Presumably the Haskell program won't like it very much if the spawned
program disables O_NONBLOCK itself?
Is there a reason (other than performance) why the use of O_NONBLOCK
can't be replaced with calling select/poll before accessing the
descriptor?
Is non-blocking I/O only necessary if you need explicit concurrency?
Or do the runtime or critical libraries require it?
--
Glynn Clements