Epoll via IOCP - Impossible?

I was looking at a Cygwin report about epoll issues and there was a recommendation from one of the devs to look at implementing it with native Windows mechanisms. So I looked at how IO Completion Ports work and, well, it's either impossible or not very easy since the IOCP architecture would require the library to aggressively pre-read data into buffers to serve up via a potential polling call, if I'm reading this all right. Seems unfortunate. Title link is a pros and cons of the two systems and a seeming confirmation of my analysis.

A Bounty of Netware drivers

Found a website that has an abundance of Netware drivers, and poking around a little, tons of nice old utilities for the DOS/Win 3.1/OS/2 era. Give it a look.

HP-UX is officially End of Life

The UNIX wars continue to come to a slow and sad end.

A brief history of TPF

A really good walkthrough of how IBM invented ACP for the airlines, then generalized it into TPF to make a ton of money on high volume transaction clients.

Galois Fields for programmers

A little spooky math for your Halloween, if you want to up your game. Particularly useful for cryptograpy.

Compiler backdoors!

If you remember the famous Ken Thompson Turing Award lecture where he demonstrated how easy it is to backdoor a C compiler, we have here the full code and an in depth look at how it works.

Volume 4, Fascicle 7

A new The Art of Computer Programming fascicle has just been released - let us begin to study it.