How difficult would it be for an x86 chip vendor to build in support for layered memory protection a/la VMS? Something that would support the separation of address space between privileged and nonprivileged executable areas? The old Vax had it - you were never in a position where you could modify the system kernel or a device driver from a program running in user mode without heavy, explicit permissions. If you tried to execute code from data, any data, forget it -- the instructions just wouldn't execute in that address space. Granted those machines blew up for other reasons, but they were secure. Close cooperation between O/S and CPU designers used to be the rule.
It seems that software people always look to a software solution, and never blame the invisible hardware. What is the extent of the communication between MS and Intel? Is there a (be polite folks) hardware solution to the O/S security question?
Check your history -- medieval history. Drawn and quartered was much worse than being stoned to death. Particularly the "drawn" part, which involved hooks.
Did something similar myself, in the days before PC's. I was working on a General Automation GA16 / 440, a.k.a "Jumboga". (You reading this, Jackson?). System was a huge amount of Fortran 66, and the mini had only a tiny 4-digit hex display on the console face. I wrote a tiny real-time program thread that would send a string sequence "FEED FACE DEAD FOOD DEDO DODO CACA" to repeat for 30 seconds a 2am on alternate Wednesdays. Was a shame the boss caught it.
Hmm... was there anything else I could have said in 4 hex digits?
I might also have wished, ideally, for a nod to the astrolabe, somewhere there, within a listing of the most incredibly nifty gadgets of all time.
For all that, it was basically a timepiece. You shot a star with the Alidade, read the angle, moved the corresponding star pointer on the rete until it matched the almucanther circle marked at the degrees you read off the alidade, then read the time off the index on the rete against the outer face. It was a sirius mater to get things right. Great use of projective geometry, the planisphere. Still trying to build one for the southern hemisphere.
That is definitely part of the problem. There are simply too many laws with too many things in each law
There should be three houses of Congress; The Senate, the House of Representatives and the Board of Editors. The third house would be comprised of disenfranchised magazine editors whose sole and entire purpose was to repeal legislation the other two houses dreamed up.
Which RDBMSes that you know are 100% ACID compliant? Please name any you know.
A very close contender would be my old friend the (now Oracle-branded) RDB from now-defunct Digital, running on VMS. It's still in use after decades of people hitting it with very large hammers, and with the less-reliable hardware of yesteryear it went through an awful lot of refinement over time. Ten years in that shop, after the first year of it's introduction we didn't lose a single committed transaction from then on. Nothing is 100% reliable but you could definitely stick a few nines after the 99 percent mark on that one. Might even be cheap nowdays, despite the Oracle and HP branding
But you know the adage -- if it works, it's obsolete.
Good point. They can't control the content, but they can control the carriage. Protocol bits are the difference between being able to stop traffic and inspect the contents of each truck vs. blocking a lane because of unsafe conditions.
We have about 16,000 workstations to control. That's 16,000 empowered individuals or 16,000 zombie drones, depending on your point of view. If we treat all 16k users as drones, we're turfed out as unresponsive. If we empower everyone, we'd need another 10k support staff and we'd be turfed out as being too expensive, even if we can find enough trained elbonians to outsource it to. The compromise position involves an awful lot of work and there's a huge market in the control systems to do that kind of management. So, guys, where is it?
This is enterprise-grade SOE territory, and is pretty much owned by Microsoft at the moment. What I'm trying to say is that if Linux is to take its fair share of corporate SOE the corporate perception of the supportability of product has to change. The way to their hearts and minds is a strongly marketed solution by people with enough clout to convince the executives they're getting a better deal. Not hype, results. You can't sell them an army of OSS volunteers, but you could sell them a Linux-focused support environment if it was comprehensive enough. They will not focus on one single implementation, they want flexibility as well as low costs. They have money to spend, generally, and they can usually pin down the "overall costs" down to the exact dime. What they want to buy is fewer headaches. Security is only one aspect, a cost they can control. Managing 16,000 custom installations is one they can not.
The big firms will embrace Linux on the desktop when they can see network deployability and end-user configuration lockdown in an easy-to-buy solution. It's a pretty major acceptance criterion. Anybody focusing on that?
(slightly off-topic, apologies) I read a short SF story a long time ago about people who had genetically re-engineered pigs so they didn't have a cloven hoof. IIRC The plot revolved around whether the resulting bacon was kosher or not, and whether or not it could be patented, and whether or not a commercial entity could own the only source of a population segment's (newly) preferred food.
If you think about it, this would be an extraordinarily contentious issue for a major segment of the population.
This made me wonder -- how much of the controversy about GE foods is based in science, and how much based in culture?
Years in pre-sales tell me there is one heck of a good use for it. Head for any customer site to set up a one-week proof of concept. Every time I've done this I've had to fight for space to set up the servers and another fight to make it half-way decent looking so we could make a good impression on the board room golfers. Setup and take down time, when engineers have to be their own roadies. Frankly, I would have killed for one.
Yes. My current contract is in a major Australian bank, where the SOE is locked down to buggery. Users can't even see their C: drive. It's there to hold software, not data, and a lot of cluey full-timers are employed to keep it that way.
The way to change a corporate computing environment is to control the default options. Whatever's easier is what people will tend to use. Whatever's easiest to support will be made the most convenient option for users. Want control? Stay on the server side...
I know of no more clear-cut example of the difference between proprietary and open source than the contrast between Encarta and Wikipedia.
Wikipedia is open to anyone who wishes to contribute and gosh, it still works. Content is added constantly and crud is scrubbed off by people who care, sort of like a child growing. Professionally I'm embedded in the Microsoft camp (what's there to fix in the Apple environment after all? And Linux is too much fun -- I get distracted) but I never use Encarta, and I'm constantly referring to Wiki for business and pleasure.
I don't really worry about Microsoft and the water economy -- Carly's treatment of two fine old technology firms show even the biggest and best companies can fall in time. In the long view, the fifty-to-100 year view, what's going to remain in use?
Would be nice if you had a button always on the screen in some discrete corner labeled "Do not disturb" to turn off reminders, popups, bells, clicks, animations and network access (yes, that too). Would be nice to have it all under control again without all the hassle of unplugging things. Sometimes you just want to write, design, and compute.
Slashdot Mirror
Yep, it's grown as a result. Umm lessee, says His Holiness. This half of the moon belongs to Spain, this half to Portugal...
It seems that software people always look to a software solution, and never blame the invisible hardware. What is the extent of the communication between MS and Intel? Is there a (be polite folks) hardware solution to the O/S security question?
Pars cantandi pars saltandi Et in bracas pars bullarum
I'm sure the Romans understood seltzer.
Check your history -- medieval history. Drawn and quartered was much worse than being stoned to death. Particularly the "drawn" part, which involved hooks.
Hmm... was there anything else I could have said in 4 hex digits?
"Hey -- come back, I'm leading you!"
Can't forget to mention UNSPSC, the UN controlled standard that manages the taxonomy behind the bar code you see on each can of alphabet soup...
For all that, it was basically a timepiece. You shot a star with the Alidade, read the angle, moved the corresponding star pointer on the rete until it matched the almucanther circle marked at the degrees you read off the alidade, then read the time off the index on the rete against the outer face. It was a sirius mater to get things right. Great use of projective geometry, the planisphere. Still trying to build one for the southern hemisphere.
And the oldest digital calculator too.
There should be three houses of Congress; The Senate, the House of Representatives and the Board of Editors. The third house would be comprised of disenfranchised magazine editors whose sole and entire purpose was to repeal legislation the other two houses dreamed up.
A very close contender would be my old friend the (now Oracle-branded) RDB from now-defunct Digital, running on VMS. It's still in use after decades of people hitting it with very large hammers, and with the less-reliable hardware of yesteryear it went through an awful lot of refinement over time. Ten years in that shop, after the first year of it's introduction we didn't lose a single committed transaction from then on. Nothing is 100% reliable but you could definitely stick a few nines after the 99 percent mark on that one. Might even be cheap nowdays, despite the Oracle and HP branding
But you know the adage -- if it works, it's obsolete.
Getting past all the rye humour, I'd be glad to pumpernickle into the scheme.
Sorry, could you clarify that please Carly?
Good point. They can't control the content, but they can control the carriage. Protocol bits are the difference between being able to stop traffic and inspect the contents of each truck vs. blocking a lane because of unsafe conditions.
This is enterprise-grade SOE territory, and is pretty much owned by Microsoft at the moment. What I'm trying to say is that if Linux is to take its fair share of corporate SOE the corporate perception of the supportability of product has to change. The way to their hearts and minds is a strongly marketed solution by people with enough clout to convince the executives they're getting a better deal. Not hype, results. You can't sell them an army of OSS volunteers, but you could sell them a Linux-focused support environment if it was comprehensive enough. They will not focus on one single implementation, they want flexibility as well as low costs. They have money to spend, generally, and they can usually pin down the "overall costs" down to the exact dime. What they want to buy is fewer headaches. Security is only one aspect, a cost they can control. Managing 16,000 custom installations is one they can not.
Build it and you can sell it.
Check one of the Pournelle links -- I think his lady developed something for reading disabilities. Read one of his columns and scan for an email link.
The big firms will embrace Linux on the desktop when they can see network deployability and end-user configuration lockdown in an easy-to-buy solution. It's a pretty major acceptance criterion. Anybody focusing on that?
If you think about it, this would be an extraordinarily contentious issue for a major segment of the population.
This made me wonder -- how much of the controversy about GE foods is based in science, and how much based in culture?
Years in pre-sales tell me there is one heck of a good use for it. Head for any customer site to set up a one-week proof of concept. Every time I've done this I've had to fight for space to set up the servers and another fight to make it half-way decent looking so we could make a good impression on the board room golfers. Setup and take down time, when engineers have to be their own roadies. Frankly, I would have killed for one.
The way to change a corporate computing environment is to control the default options. Whatever's easier is what people will tend to use. Whatever's easiest to support will be made the most convenient option for users. Want control? Stay on the server side...
How to let your enemy win, in one easy lesson -- become them.
Not to mention 4,500 articles (Encarta) vs. 471,775 (Wikipedia, qty probably already out of date)
Wikipedia is open to anyone who wishes to contribute and gosh, it still works. Content is added constantly and crud is scrubbed off by people who care, sort of like a child growing. Professionally I'm embedded in the Microsoft camp (what's there to fix in the Apple environment after all? And Linux is too much fun -- I get distracted) but I never use Encarta, and I'm constantly referring to Wiki for business and pleasure.
I don't really worry about Microsoft and the water economy -- Carly's treatment of two fine old technology firms show even the biggest and best companies can fall in time. In the long view, the fifty-to-100 year view, what's going to remain in use?
Bonch, you're fired.
Would be nice if you had a button always on the screen in some discrete corner labeled "Do not disturb" to turn off reminders, popups, bells, clicks, animations and network access (yes, that too). Would be nice to have it all under control again without all the hassle of unplugging things. Sometimes you just want to write, design, and compute.