|
|
|
|
|
|
Jon suggested that networks shouldn't be treated as devices, and shouldn't need to be explicitly programmed. In otherwords they should transparently move data from place to place similarly to the computer's backplane.
Reacting to the comments of Margo Seltzer (and others) earlier in the day, Geoff (left) argued that "databases should not drive operating systems. Databases are a single very specialized applications. If we start bending over backwards to support databases we're going to forget the fact that we're not writing special purpose operating systems."
John observed that lots of people at the workshop were using Pilot pen-based computers. He commented that carrying them around, and pulling them out all the time is a pain in the neck. If people want personal computers that they can carry around with them, we should find a better place to carry them around. "Where is a better place?" John asked. "The answer is, give up an eye! Think about it, you get a direct connection to the optical nerve. The natural question then is, are you willing to give up an eye for 100 megabits?"An unknown attendee responded, "Why not? We've already given up two carpal tunnels for emacs!"
Margo Seltzer wanted to know if this is would be called "optical computing".
Brian read the text of a bet that they made at the 1993 HotOS: "Bet: Brian says -
By Dec. 31, 1996, Sun Microsystems will no longer be doing new development on a UNIX-like operating system.
Loser gives winner 1 bottle of Stag's Leap Vineyards most recent release of C abernet Sauvignon (best available quality)."
At this point, Brian handed Jeff the aforementioned bottle."Now here's the really stupid part," Brian continued as he finished reading. "Bet is void if Sun goes out of business."
Undetered by his loss, Brian agreed to a new bet with Jeff, that by Dec. 31, 2001, Digital Equipment Corportation will no longer be in business.
Margo Seltzer, eager to win her own bottle of wine, asked Brian if he'd also like to bet that Harvard University will be out of business by 2001.
"By the year 2001," Sape(left) inquired, "will we be terribly ashamed because we are not working for Microsoft, because everyone will believe that we are not good enough?" Jeff Mogul responded by proposing that the attendees form a tontine, to be paid to the last person to go work for Microsoft.
Geoff Keunning observed that this was exactly the *opposite* of what had been happening so far.
Fred presented a list of the top ten themes he anticipates for
HotOS-VII:
10: JavaOS
9: Randomized Operating Systems
8: WebOS
7: Why JavaOS failed
6: The electron is the computer
5: Zero cost IPC
4: Spinning drunkenly in a pool of processors,
an existential tale of flux.
3: Foreign access fees for ATM
2: MS-DOS ("and we all know who the MS stands for...
Martha Stewart!)
1: The rise of monolithic kernels.
David was introduced by David Cohen of Notre Dame who pointed out the historical significance of the first two papers of the workshop. "It was the last time any of us will hear a paper about a full operating system designed by anybody but Microsoft."David (Steere) also presented a top ten list. His was the top ten careers for ex-operating systems researchers:
10: Start an ISP 9: Off-shore electronic money laundering 8: Assistant to Martha Stewart 7: Join Apple for their lucrative severance package 6: Stoop to teaching 5: (with credit to Jeff) Collect the tontine. 4: Exploit security holes in Active X 3: Sell out to Microsoft, such as by becoming Rick Rashid's personal assistant. 2: Crash test dummy for ethernet collisions 1: COBOL hacking for the year 2000
Mark(right)'s opinion was motivated by the fact that on several occasions that day he had heard people say of their systems that it was a "perfectly reasonable thinkg for programmers to use." Mark was willing to bet that nobody had actually sat real programmers in front of their systems and asked them to use it. "All I want is for people to test their supposed-to-be-easy-to-use things on real people in order to verify that they are easy to use," Mark said. 
John(left) noted that there is a big push in industry toward "client-server" computing. "I think this is basically bullshit because the servers are overpriced," John said. He claimed that the hardware manufacturers are taking advantage of massive markups on their servers, basically selling clients at cost and making all of their profits on servers. John argued that this situation "will turn around and bite anybody who expects to be making all of their money on servers, because just like the PCs ate the workstations, the clients will eat the servers."
Jay Lepreau called this, "the demise of little iron and the rise of tin."
Warning that his opinion might seem like "obvious flame bait," Anil claimed that Lunix is the future of operating system research and development. His argument went as follows. In three or four years, Microsoft will have NT running on everyone's desk tops. After that, NT will never leave people's machines, and desktop operating systems (i.e., NT) won't change much, except for bug fixes. This will give Linux time to adapt and develop its WIN32 API. The Linus will co-opt NT because it will run all of the same applications, only better.