Linux for Two Cans Strung Together, v.0.0.1a
Fake News written by on Wednesday, October 21, 1998
Dateline: a secret location somewhere in the Arizona desert
Just when you thought that Linux couldn't possibly be ported to another platform, scientists describing themselves as "attached to a private, secret, not-for-profit research institute" have announced the leanest Linux version to date, and possibly ever.
"It started out as a lark, a can-it-be-done sort of thing," recalls Dr. David F. Schmierck, who is currently helping to beta-test the new port. "We're still getting over the shock of our initial success."
No, the new port doesn't run on some fancy high-speed RISC chip, or an older microcomputer like the Atari; in fact, it doesn't run on a microprocessor at all.
"So far the architecture we've had the most success with is a matched pair of Dinty Moore Stew cans, I think they're listed as 'size number ten,'" explains Schmierck. "The current networking medium of choice is bailing twine."
Yes, folks, you heard it here first: Linux has been successfully ported to a pair of tin cans with a string between them.
Of course, the project is still in its early stages, and Schmierck describes the current state of the code as "heavily alpha."
"Obviously, this architecture imposes some serious limitations, some of which may prove insurmountable," he confides. For example, while the current version preserves some of UNIX's trademark multi-user format, it only works if the number of users is precisely two; try it with zero, one, or three login sessions, and the system immediately freezes or becomes otherwise unusable. Stability over time has also been problematic, said Schmierck. "The guy on the other end usually starts to lose interest after about ten or fifteen minutes," he said, "although we have recorded uptimes in the three-hour range under light system load with experienced UNIX administrators."
Still, the research team is optomistic about the project's future, and one tester described the general mood as "a little giddy" over their success so far. Currently planned improvements include further miniaturization -- Dr. Schmierck believes that the system will probably work on a pair of Campbell's Soup cans with only minor modifications -- as well as dreaming up some way around the two user limit, something which the team unanimously agrees will be vital if "Grover clusters" are to achieve wider acceptance.
Like the rest of the contemporary Linux world, the Grover team recognizes that applications will be a key element in their project's future or lack thereof. So, what can Grover clusters be used for? Here, too, the story so far has been mixed.
Says Schmierck, "One area where we've had unqualified success has been in voice recognition. Grover clusters are better than any platform I know of when it comes to that. Voice over IP has also been quite good. But other forms of heavy I/O have been almost uniformly disastrous. Try to transmit a jpeg, and just watch the whole thing grind to a halt." Attempts to run the Apache web server have also been so far unsuccessful, a situation which Dr. Schmierck says is independent of the two-user problem: "Background processes still have us stumped."
Asked what kind of initial reaction they've seen from the larger Open Source community, the Grover team puts on a brave face. "This is a pretty radical departure, and stuff like this can take a while to catch on. So far there hasn't really been much interest, and one Linux distributor actually hung up the phone when we asked them to mirror our ftp site. But if some punk from Finland can start from Minix and make it useful, we figure we can't lose."