<> Well, there were some recent modifications to this portion of the code to
<> solve some bugs, so it's possible that those bugs were covering over this
<> problem, if people were using writeable _non-blocking_ sockets.
I /am/ using a non-blocking socket, btw.
I figured, that there must have been a "conceptual" problem after I went
through all permutations of using O_NDELAY, O_NONBLOCKING and FIONBIO and
still wound up either being stuck in a syswrite() or have it returning
EGAIN in /very/ short succession after being repeatedly being called-back
by Event ...
-c
