Bart, Thanks for pointing out those bugs. Somebody actually reported the problem with "accept" about a month ago and the fix will be included in the next release. It actually turned out to be very simple: I'd failed to initialise the size of the buffer to receive the address before calling "accept". Let me know if you'd like more details of the fix.

I hadn't noticed the bug in Socket.Ctl.setNBIO . I'll include that fix in the next release as well.

David

On Friday, June 28, 2002, at 10:19 , Bart Jacobs wrote:

When I use Socket.accept in Poly/ML 4.1.2 on Windows, I get SysErr("accept failed", SOME EFAULT).

Also, I think I found a bug in the driver source code, but I'm not sure that the latter causes the former. Specifically, in the code for Socket.Ctl.setNBIO (POLY_SYS_net number 43), it says

ioBits |= ~NON_BLOCKING_IO

instead of

ioBits |= NON_BLOCKING_IO