However, if you leave your door open and some kid takes the gun on the coffee table to kill a few people at the school, you should expect a few civil suits to be thrown in your direction and to be legally responsible (negligence). If you were not negligent (ie, locked the fucking door), it's unfortunate, but not your fault legally.
It's a similar deal with wireless LANs. If some luser comes on to your unencrypted network and uses it to deface microsoft.com, it's going to be very hard to show that you weren't negligent when Steve "Monkey-Boy" Ballmer decides that they need to make an example of somebody. If it's encrypted, you are as much a victim as Microsoft, since your industry-standard security measures were defeated. If you were a negligent bastard and didn't use the industry-standard security measures, much like putting your gun behind a lock -- even if that lock could be easily defeated -- you are not the victim. I have the flu and am not thinking entirely clearly, so I apologive that I'm not especially coherent, but do you understand what I was saying now?
WEP is not useful for anything than discouraging the casual bandwidth leech, if that matters to you at all.
WEP may be useful in one other way -- it gives you some legal protection if someone else uses your wireless network to do something malicious. Running your network unencrypted could be seen as the equivalent of leaving your front door open when you're not home.
I actually thought it was more NeXTish than Warpish, meself. (I was a Win9x beta tester and also used NeXTstep and OS/2, and shortly later started to use Linux... but I still miss Netware 4.1.)
Based on what I've read, another problem that Apple had getting Copland out the door was that nobody (good) wanted to work on it. The other projects going on were much cooler.
Actually, what I remember of memory management in traditional MacOS is extremely innovative -- and one of the main reasons that it was nearly impossible to implement protected memory *without breaking compatibility*.
Apple has tried a few times before MacOS X (ie, buying the technology from NeXT) to reimplement the Macintosh -- the "Pink" project, using IBM's variant of the Mach kernel to implement a whole new OS, Copland -- the ill-fated attempt to create a new MacOS that was supposed to be 8.0, and more that never made it out of the ATG (while it still existed).
The memory management in the MacOS was extremely, uh, unique and uh, special. Rather than allocating a direct pointer to memory, you allocated a handle. A handle was a pointer to a pointer. At the top of the heap was a block of handles (32-bit pointers), and they pointed to other locations in the heap (actual memory locations).
That allowed the MacOS to defragment memory. If an app or the system allocated a larger block of memory than was available in the heap, the system's memory manager would shuffle around the contents of the heap (updating the handles) and try to create a larger free area.
Since the original hardware didn't have a proper MMU (no virtual memory or paging), this allowed much lower memory requirements for complex apps that would need to allocate large, contiguous memory spaces.
One unfortunate side effect was having to set the memory partition size of an application before running it (either under MultiFinder in 6.x or under the multitasking finder of 7.x). It's another place where the designers did something extremely cool... that just didn't work well with multiple applications running.
So, after this long (and probably at least partially incorrect) rant, all I'm really trying to say is be nice to Apple. They showed the rest of the world what *not* to do.:) Just because it didn't work out well does not mean is was not innnovative.
Incorrect. For parity, you need one additional chip (in the example you cite.) So, for every 8 bits, you need one additional bit. All that parity does is tell you that it failed to store the incorrect information. ECC is one step further -- think of it like RAID-5 for memory. It not only tells you that it's incorrect, it lets you determine the correct value. IIRC, for 32-bit memory you need 40 bits for parity and 36-bits for ECC. It may be 42 bits for ECC.
Here's what I don't get about these discussions about why somebody "shouldn't" work on something like Twin -- why does it matter? Someone who wants a pretty gui is not going to use this. I really can't think of anyone who will (other than the people programming it and a few old DesqView junkies), but why does it prevent you or anyone else from using X11/Gnome or X11/KDE?
Open source isn't about taking over the world, not for most of the people who actually contribute to it (as opposed to the leeches who just complain that "But this won't help kill M$!"). It's about scratching an itch.
There are a few things that I'm working on for myself that I'm not sure anybody else would even care about. I can't think of many things that would piss me off more than someone telling me that I shouldn't waste my time on that, since it wouldn't help them. The only times that you're allowed to do that is if a) you are my wife, b) you are my employer (and I'm at work), or c) you are paying me to program. Beyond that, it's none of your fucking business. Don't like it? Don't use it. Like it, but think it needs improvement (documentation, etc)? Join in and help.
1) Forget the 3800 -- get the 4800. cPCI is a dead-end from Sun. (Trust me - I admin a 3800. Couldn't even get GigE for a long time after we had the serengeti).
2) Fuck replacing a faulty NIC -- Netware can do that on some Compaq and other servers with Hotplug PCI. I can replace a fucking processor in a running system without a fucking reboot. Hell yeah.
More likely, Sun fucked Apple in some way -- some snub somewhere that was enough to piss off Jobs. Between Ellison (CEO of Oracle, on Apple's board, insane), Jobs (insane), and McNealy (again, insane), there's more than enough ego to fuck anything up.
Incorrect. NeXTstep/OpenStep did not have a limit on the number of platforms that a binary could be fat for, and in fact ran on sparc32, hppa, x86, 68K, and I think there were ports to ppc and 88K as well (possibly not released.)
The binaries were built automagically by Project Builder using cross-compilers, and were stored in the application directory, so an application (that often didn't need to be "installed") would have a structure like Application.app/resources, Application.app/Application.next-mach-x86, Application.app/Application.sun-openstep-sparc -- store the contents of the application, everything it needed, as this single bundle that just appeared as a single icon in the file browser.
OS X does preserve this architecture for native apps. OpenStep also didn't need to run as a full operating system -- it was available as a set of libraries and extensions to Windows NT, for example.
I'm tired. Please correct any asinine mistakes I made, and please don't correct spelling or grammar.
I haven't done anything resembling objective testing yet, but the NFS client *feels* a lot faster in 9 than in 8 or 7.x. Large directory listings in particular are faster, and file transfers seem faster (again, I apologize for claiming this without numbers, but I haven't had time to really work it over.)
Re:I don't know the answer, but don't use "and"!
on
Eleventy What?
·
· Score: 1
I reserve my personal hatred for people who say "Barnes and Nobles." I only make one exception -- people from Chicago can't help themselves. Wrigley's Field, indeed.
Grocery stores sell them already. They call them "freezer bags.":)
Re:Why do some many prefer Gnome then ?
on
Has GNOME Become LAME?
·
· Score: 2, Interesting
KDE is updated more often (major revs) as well, so it's not a clear statistic. It could just be that more people emerge KDE to jump versions than Gnome users (which have been at 2.0 for a while).
Incorrect. The 550 Mhz USII mentioned is in a Blade 100 or 150, which means that it's fucking slow. It's a IIe, not a II. A PIII/550 would smoke it for web serving.
256K of cache on die, ALI chipset board that's a lot like a PC, slow PC133 (with very high latency) memory, dog-fucking-slow disk, unless they're using SCSI.
Slashdot Mirror
It's a similar deal with wireless LANs. If some luser comes on to your unencrypted network and uses it to deface microsoft.com, it's going to be very hard to show that you weren't negligent when Steve "Monkey-Boy" Ballmer decides that they need to make an example of somebody. If it's encrypted, you are as much a victim as Microsoft, since your industry-standard security measures were defeated. If you were a negligent bastard and didn't use the industry-standard security measures, much like putting your gun behind a lock -- even if that lock could be easily defeated -- you are not the victim. I have the flu and am not thinking entirely clearly, so I apologive that I'm not especially coherent, but do you understand what I was saying now?
Every country has different rules...
Then the code will be release in three years after being fully audited...
I actually thought it was more NeXTish than Warpish, meself. (I was a Win9x beta tester and also used NeXTstep and OS/2, and shortly later started to use Linux... but I still miss Netware 4.1.)
Based on what I've read, another problem that Apple had getting Copland out the door was that nobody (good) wanted to work on it. The other projects going on were much cooler.
Apple has tried a few times before MacOS X (ie, buying the technology from NeXT) to reimplement the Macintosh -- the "Pink" project, using IBM's variant of the Mach kernel to implement a whole new OS, Copland -- the ill-fated attempt to create a new MacOS that was supposed to be 8.0, and more that never made it out of the ATG (while it still existed).
The memory management in the MacOS was extremely, uh, unique and uh, special. Rather than allocating a direct pointer to memory, you allocated a handle. A handle was a pointer to a pointer. At the top of the heap was a block of handles (32-bit pointers), and they pointed to other locations in the heap (actual memory locations).
That allowed the MacOS to defragment memory. If an app or the system allocated a larger block of memory than was available in the heap, the system's memory manager would shuffle around the contents of the heap (updating the handles) and try to create a larger free area.
Since the original hardware didn't have a proper MMU (no virtual memory or paging), this allowed much lower memory requirements for complex apps that would need to allocate large, contiguous memory spaces.
One unfortunate side effect was having to set the memory partition size of an application before running it (either under MultiFinder in 6.x or under the multitasking finder of 7.x). It's another place where the designers did something extremely cool... that just didn't work well with multiple applications running.
So, after this long (and probably at least partially incorrect) rant, all I'm really trying to say is be nice to Apple. They showed the rest of the world what *not* to do. :) Just because it didn't work out well does not mean is was not innnovative.
Incorrect. For parity, you need one additional chip (in the example you cite.) So, for every 8 bits, you need one additional bit. All that parity does is tell you that it failed to store the incorrect information. ECC is one step further -- think of it like RAID-5 for memory. It not only tells you that it's incorrect, it lets you determine the correct value. IIRC, for 32-bit memory you need 40 bits for parity and 36-bits for ECC. It may be 42 bits for ECC.
Open source isn't about taking over the world, not for most of the people who actually contribute to it (as opposed to the leeches who just complain that "But this won't help kill M$!"). It's about scratching an itch.
There are a few things that I'm working on for myself that I'm not sure anybody else would even care about. I can't think of many things that would piss me off more than someone telling me that I shouldn't waste my time on that, since it wouldn't help them. The only times that you're allowed to do that is if a) you are my wife, b) you are my employer (and I'm at work), or c) you are paying me to program. Beyond that, it's none of your fucking business. Don't like it? Don't use it. Like it, but think it needs improvement (documentation, etc)? Join in and help.
2) Fuck replacing a faulty NIC -- Netware can do that on some Compaq and other servers with Hotplug PCI. I can replace a fucking processor in a running system without a fucking reboot. Hell yeah.
More likely, Sun fucked Apple in some way -- some snub somewhere that was enough to piss off Jobs. Between Ellison (CEO of Oracle, on Apple's board, insane), Jobs (insane), and McNealy (again, insane), there's more than enough ego to fuck anything up.
The binaries were built automagically by Project Builder using cross-compilers, and were stored in the application directory, so an application (that often didn't need to be "installed") would have a structure like Application.app/resources, Application.app/Application.next-mach-x86, Application.app/Application.sun-openstep-sparc -- store the contents of the application, everything it needed, as this single bundle that just appeared as a single icon in the file browser.
OS X does preserve this architecture for native apps. OpenStep also didn't need to run as a full operating system -- it was available as a set of libraries and extensions to Windows NT, for example.
I'm tired. Please correct any asinine mistakes I made, and please don't correct spelling or grammar.
You've obviously never used a DECwriter. Those keyboards were horrible.
I haven't done anything resembling objective testing yet, but the NFS client *feels* a lot faster in 9 than in 8 or 7.x. Large directory listings in particular are faster, and file transfers seem faster (again, I apologize for claiming this without numbers, but I haven't had time to really work it over.)
I reserve my personal hatred for people who say "Barnes and Nobles." I only make one exception -- people from Chicago can't help themselves. Wrigley's Field, indeed.
To hell with that! 0xDECAFBAD is my choice.
Grocery stores sell them already. They call them "freezer bags." :)
KDE is updated more often (major revs) as well, so it's not a clear statistic. It could just be that more people emerge KDE to jump versions than Gnome users (which have been at 2.0 for a while).
Statistics lie.
What's wrong with 'sudo bash'? (further off topic...)
Too often, the Supreme Court forgets the Ninth. It's been interpreted away to zilch.
Enjoy the ride.
ITYM "metric arse-load."
256K of cache on die, ALI chipset board that's a lot like a PC, slow PC133 (with very high latency) memory, dog-fucking-slow disk, unless they're using SCSI.
This is not your father's E450.