io.backend.unix: Try again when we get ENOBUFS

My understanding is the tcp buffers can fill up and we get notified
with ENOBUFS. After some time they should be free again.

This makes the parallel-map read more reliable

related to #2771

diff --git a/basis/io/backend/unix/unix.factor b/basis/io/backend/unix/unix.factor
index 1b2cbc4455df4d28353631dd4d2e4761635c2260..f58e2d942119f945546af85a780bf1df76bb7671 100644 (file)
--- a/basis/io/backend/unix/unix.factor
+++ b/basis/io/backend/unix/unix.factor
@@ -109,6 +109,7 @@ M: fd drain
         errno {
             { EINTR [ 2drop +retry+ ] }
             { EAGAIN [ 2drop +output+ ] }
+            { ENOBUFS [ 2drop +output+ ] }
             [ (throw-errno) ]
         } case
     ] if ;