>... the Direct Marketing Association, or as I like to call them "the first horseman of the apocalypse."
War? I dunno, Pestilence seems more appropriate to me:) Tho the biblical passage seems more to have five: Antichrist, War, Famine, then Death/Pestilence and Hell making a joint appearance, so I suppose you're not too far off after all. And the USPS IS GIVING IT ITS CROWN, AND IT WILL GO OUT AND CONQUER, SPAM IS THE HARBINGER OF END TIMES REPENT REPENT REPENT ok theyre saying i have to take my meds now bye
> One thing though: how come he pronounces Linux as/LIN-ucks/, but Linus as "LIE-nus"?
American pronunciation. "LIN-ucks" is the winner in the pronunciation debate, probably because it's closest to how Linus pronounces it when his accent isn't as pronounced. In his native tongue, his name is "LEE-noose" and the OS is "LEE-nooks". He didn't name it anyway, he called it Freeix.
Short answer: A SunRay is a dumb terminal with a full graphic framebuffer instead of just a character display.
A SunRay has a framebuffer (video card) in it, 100 megabit ethernet, and a 4 port USB hub through which the keyboard and mouse connect. It has a smartcard slot in the front. They're totally dumb terminals, the session runs on the SunRay server machine, typically a E4500, just as if they were local X sessions run on virtual consoles. Uses standard X, but the video driver portion of the X server redirects over the network. The SunRays don't have IP addresses, the protocol is just above ethernet, so the SunRay terminals exist on a private network.
The smartcard slot lets you do something called "hot-desking", where your virtual framebuffer is attached to your card ID. Without a card, it's attached to the Ethernet address of the SunRay, with the card it's attached to the card, so if you yank it out, it disconnects the fb and resets the sunray back to the xdm login screen. Stick your card in another sunray and it puts your desktop up there in the exact state it was in, even the mouse cursor is in the same place. No one logs out of a sunray or even locks their screen, they just yank their card.
What's really amazing about it is that if you spend an equivalent amount on hardware for the SunRay server and terminals that you would on workstations, they are actually *fast*. There are no visual artifacts from even full screen redraws, no lag whatsoever -- except if a process goes berzerk and you don't have good memory limits set (I think SRM in solaris 8 handles that pretty well now tho). And they're just zero administration, you have one server to manage and that's it. You just have to be more careful with that one server.
They're not for everyone -- engineers may need to fiddle with local configuration for example, and probably need private resources for intensive compiles, but for the majority of office drones, they're just fine. They're definitely not the abomination the JavaStation was.
> Notice that the author is conspicuous in comparing apples to oranges (GNU libc to BSD libc)
God forbid one compare one libc against another! Oh that's right, glibc is far more bloated and monolithic. I understand BeOS renames it to libunix instead. So much for unix's vaunted orthogonality. And it gets worse and worse every release. Drepper and company shoehorn every last function they can grab into glibc, and it all has the look of a power grab -- god forbid the unwashed should have any alternatives but our flawless implementation in glibc.
libc was an ill-conceived idea from the start, and it shows every sign of getting worse with glibc.
> The Good Thing is that it's network transparent, so vendors can use it to do a "deploy applications over the web" strategy (or devices like the SUN Ray).
Try again. SunRays (nee Corona nee NewT) redirect the whole framebuffer across the network. Think VNC in hardware.
There's no game there, but there's two isometric tile engines for WorldForge, namely uclient and anvil as well as a couple 3d clients too. Free for anyone to use for their own games (they have no game-specific code in them, WF is more of a toolbox for RPGs, with a med/fan world as its flagship implementation). And the art is better than anything else I've seen in an open-source game.
<plug>WF could always use artists, writers, graphics and AI coders, etc. Go on over there and check it out.</plug>
All this Alice xor Bob = Secret scheme gets you is Alice and Bob both named equally culpable co-defendants. Is any judge or jury going to believe that Alice's data was just random and by some amazing coincidence, turned out to be the Secret when XOR'd with some other data? You can do a doubly linked list with a single pointer per node and xor tricks, and far from obscuring both nodes, the same data item reveals both pointers. By the time you're in court, you better have a stronger case than Alice's (and against Scientology, you have mountains of case law on your side)
Better control over locking than in Java - I hate the way they do synchronised methods. I much prefer the control of placing mutex locks and unlocks where I want them - I can get it more fine grained, and it's much more obvious what I'm doing (and more consistent with what experienced programmers are used to seeing).
You can do exactly this by creating a Mutex object (you can even pass it around on the stack and create it anonymously) and then using a synchronized block. I suggest getting a copy of Java Threads, by ORA.
And comparing COM v. CORBA is the most mismatched comparison I've ever seen. Try DCOM, which is just as much a monstrosity.
> So what? syntactical sugar, nothing more. [repeated for each point]
All languages are nothing more than syntactic sugar on top of turing machines. Did you have an actual point, or were you just scoring knee-jerk anti-Microsoft points? I give it three out of five Slashdots.
Device drivers. Malloc replacements. C interfaces. The knee-jerk factor on slashdot is unbearable on this topic, and was when I read the last article on this language. Do I have to read at threshold 5 whenever the word "Microsoft" is in an article now?
There is no 8 or 32 bit virtual machine. I don't know where you get your info. DOS has *never* been 8 bit. Its main selling point when it was released was that it was 16 bit and CP/M was not.
Hello fallacy of composition. The world is not a freakin watch, ergo there is no case for a watchmaker. I can't believe some people still buy this tired old theist argument.
I goofed up slightly. I'm told -RELEASE is a "checkpointed" -STABLE, and would be sort of the equivalent of a new distribution from a vendor such as redhat or mandrake. 4.0-RELEASE was just like any other.0 release though, and I was warned away in pretty strong language from getting it til it hit -STABLE if I didn't know what I was doing.
You don't want 4.0-RELEASE, you want the latest 4.0-STABLE. Then when you install cvsup, make sure you use tag=RELENG_4 or you'll end up getting forcibly upgraded to 5.0-CURRENT, which is *not* stable. Was not pretty when this happened to me.
comparing version number systems, BSD to Linux:
5.0-CURRENT is the equivalant of 2.3.x 5.0-RELEASE will be the equivalent of 2.4pre 5.0-STABLE will be the official equivalent of 2.4
FreeBSD does not have any kind of patchlevel number for its minor releases, since it is updated constantly in CVS, so if you stay on the edge of whatever version, be it CURRENT, RELEASE, or STABLE, your build date is part of the uname. If you install from a snapshot out of FTP, then your distribution version looks something like 4.0-STABLE-20000624 (i.e. Jun 6, 2000). When enough patches accumulate to make it a recommended upgrade for everyone, then the powers that be (presumably jordan?) bump the minor version number.
I'm a FreeBSD user, and I love/usr/ports like nothing else, but the package manager is awful. Bad as slackware, possibly worse. It doesn't approach rpm, much less apt. There isn't any way to upgrade packages, you pretty much have to remove the old one before installing the new one. Packages seem to have versioning, but the version is still part of the name, so when you update ports with cvsup, a "make deinstall" won't even find the old package built from ports. Since making a new version from a new port will overwrite the old files, I tend to just not bother with deinstalling the old version and thus pkg_info becomes a sort of history (the consistency checker can at least tell me which ones are "really" installed).
Ports is nice, I do wish it built more robust packages tho.
FreeBSD has supported >2GB files for as long as I can remember. I don't know how to tell for sure, but it would also appear FreeBSD uses a 3gig user, 1gig kernel address space, since absolute addresses well above 2G were reported available when I ran a malloc arena prospector program.
Linux support is superb. I'm posting this reply with linux netscape 4.7, and there's no detectable difference in performance, behavior, or stability. Actually it's more stable than I've ever had it under linux (probably because I'm disinclined to constantly futz with the system installed in/compat/linux)
I do have to disappoint you a bit: UFS+softupdates will not give you journalling. Soft updates is merely another block cache and deferred write queue. Speeds up the filesystem tremendously, but doesn't really add any new features.
Slashdot Mirror
> We have a double-standard for privacy issues on the internet.
... do I really need to continue?
s/Apple/Microsoft/
> ... the Direct Marketing Association, or as I like to call them "the first horseman of the apocalypse."
:) Tho the biblical passage seems more to have five: Antichrist, War, Famine, then Death/Pestilence and Hell making a joint appearance, so I suppose you're not too far off after all. And the USPS IS GIVING IT ITS CROWN, AND IT WILL GO OUT AND CONQUER, SPAM IS THE HARBINGER OF END TIMES REPENT REPENT REPENT ok theyre saying i have to take my meds now bye
War? I dunno, Pestilence seems more appropriate to me
> Or maybe you could stop being a whining pussy and start coding.
Classic Linux response. And people wonder why it's seen as "for nerds only".
Erster Pfosten!
Heiße Körner!
> One thing though: how come he pronounces Linux as /LIN-ucks/, but Linus as "LIE-nus"?
American pronunciation. "LIN-ucks" is the winner in the pronunciation debate, probably because it's closest to how Linus pronounces it when his accent isn't as pronounced. In his native tongue, his name is "LEE-noose" and the OS is "LEE-nooks". He didn't name it anyway, he called it Freeix.
> So if you look at something that's read with red glasses on, you see it as white.
Uh no, you see it as red. Will blend in completely with a red background too.
I'll take perl's $%@{[()]} soup over python's forced indentation any day. No language I'll ever write in starts by assuming I can't write good code.
Short answer: A SunRay is a dumb terminal with a full graphic framebuffer instead of just a character display.
A SunRay has a framebuffer (video card) in it, 100 megabit ethernet, and a 4 port USB hub through which the keyboard and mouse connect. It has a smartcard slot in the front. They're totally dumb terminals, the session runs on the SunRay server machine, typically a E4500, just as if they were local X sessions run on virtual consoles. Uses standard X, but the video driver portion of the X server redirects over the network. The SunRays don't have IP addresses, the protocol is just above ethernet, so the SunRay terminals exist on a private network.
The smartcard slot lets you do something called "hot-desking", where your virtual framebuffer is attached to your card ID. Without a card, it's attached to the Ethernet address of the SunRay, with the card it's attached to the card, so if you yank it out, it disconnects the fb and resets the sunray back to the xdm login screen. Stick your card in another sunray and it puts your desktop up there in the exact state it was in, even the mouse cursor is in the same place. No one logs out of a sunray or even locks their screen, they just yank their card.
What's really amazing about it is that if you spend an equivalent amount on hardware for the SunRay server and terminals that you would on workstations, they are actually *fast*. There are no visual artifacts from even full screen redraws, no lag whatsoever -- except if a process goes berzerk and you don't have good memory limits set (I think SRM in solaris 8 handles that pretty well now tho). And they're just zero administration, you have one server to manage and that's it. You just have to be more careful with that one server.
They're not for everyone -- engineers may need to fiddle with local configuration for example, and probably need private resources for intensive compiles, but for the majority of office drones, they're just fine. They're definitely not the abomination the JavaStation was.
> Notice that the author is conspicuous in comparing apples to oranges (GNU libc to BSD libc)
God forbid one compare one libc against another! Oh that's right, glibc is far more bloated and monolithic. I understand BeOS renames it to libunix instead. So much for unix's vaunted orthogonality. And it gets worse and worse every release. Drepper and company shoehorn every last function they can grab into glibc, and it all has the look of a power grab -- god forbid the unwashed should have any alternatives but our flawless implementation in glibc.
libc was an ill-conceived idea from the start, and it shows every sign of getting worse with glibc.
> The Good Thing is that it's network transparent, so vendors can use it to do a "deploy applications over the web" strategy (or devices like the SUN Ray).
Try again. SunRays (nee Corona nee NewT) redirect the whole framebuffer across the network. Think VNC in hardware.
> This is stupid. The US already has the ESRB, so we don't NEED any laws for this.
Were you aware that British Columbia is not part of the United States? You're a product of the US primary education system, aren't you?
> Dylan and Klebold (the killers at Columbine High School
Dylan Klebold was one of them. Eric Harris was the other.
There's no game there, but there's two isometric tile engines for WorldForge, namely uclient and anvil as well as a couple 3d clients too. Free for anyone to use for their own games (they have no game-specific code in them, WF is more of a toolbox for RPGs, with a med/fan world as its flagship implementation). And the art is better than anything else I've seen in an open-source game.
<plug>WF could always use artists, writers, graphics and AI coders, etc. Go on over there and check it out.</plug>
All this Alice xor Bob = Secret scheme gets you is Alice and Bob both named equally culpable co-defendants. Is any judge or jury going to believe that Alice's data was just random and by some amazing coincidence, turned out to be the Secret when XOR'd with some other data? You can do a doubly linked list with a single pointer per node and xor tricks, and far from obscuring both nodes, the same data item reveals both pointers. By the time you're in court, you better have a stronger case than Alice's (and against Scientology, you have mountains of case law on your side)
bah. i guess either netscape doesn't render , or slashdot strips it out. The first paragraph was a cite.
Better control over locking than in Java - I hate the way they do synchronised methods. I much prefer the control of placing mutex locks and unlocks where I want them - I can get it more fine grained, and it's much more obvious what I'm doing (and more consistent with what experienced programmers are used to seeing).
You can do exactly this by creating a Mutex object (you can even pass it around on the stack and create it anonymously) and then using a synchronized block. I suggest getting a copy of Java Threads, by ORA.
And comparing COM v. CORBA is the most mismatched comparison I've ever seen. Try DCOM, which is just as much a monstrosity.
> So what? syntactical sugar, nothing more.
[repeated for each point]
All languages are nothing more than syntactic sugar on top of turing machines. Did you have an actual point, or were you just scoring knee-jerk anti-Microsoft points? I give it three out of five Slashdots.
We heard this lameass CASH joke the first thousand times the first time this language was announced. Can we please give it a rest?
> When do I absolutely *HAVE* to use pointers?
Device drivers. Malloc replacements. C interfaces. The knee-jerk factor on slashdot is unbearable on this topic, and was when I read the last article on this language. Do I have to read at threshold 5 whenever the word "Microsoft" is in an article now?
There is no 8 or 32 bit virtual machine. I don't know where you get your info. DOS has *never* been 8 bit. Its main selling point when it was released was that it was 16 bit and CP/M was not.
Hello fallacy of composition. The world is not a freakin watch, ergo there is no case for a watchmaker. I can't believe some people still buy this tired old theist argument.
I goofed up slightly. I'm told -RELEASE is a "checkpointed" -STABLE, and would be sort of the equivalent of a new distribution from a vendor such as redhat or mandrake. 4.0-RELEASE was just like any other .0 release though, and I was warned away in pretty strong language from getting it til it hit -STABLE if I didn't know what I was doing.
You don't want 4.0-RELEASE, you want the latest 4.0-STABLE. Then when you install cvsup, make sure you use tag=RELENG_4 or you'll end up getting forcibly upgraded to 5.0-CURRENT, which is *not* stable. Was not pretty when this happened to me.
comparing version number systems, BSD to Linux:
5.0-CURRENT is the equivalant of 2.3.x
5.0-RELEASE will be the equivalent of 2.4pre
5.0-STABLE will be the official equivalent of 2.4
FreeBSD does not have any kind of patchlevel number for its minor releases, since it is updated constantly in CVS, so if you stay on the edge of whatever version, be it CURRENT, RELEASE, or STABLE, your build date is part of the uname. If you install from a snapshot out of FTP, then your distribution version looks something like 4.0-STABLE-20000624 (i.e. Jun 6, 2000). When enough patches accumulate to make it a recommended upgrade for everyone, then the powers that be (presumably jordan?) bump the minor version number.
I'm a FreeBSD user, and I love /usr/ports like nothing else, but the package manager is awful. Bad as slackware, possibly worse. It doesn't approach rpm, much less apt. There isn't any way to upgrade packages, you pretty much have to remove the old one before installing the new one. Packages seem to have versioning, but the version is still part of the name, so when you update ports with cvsup, a "make deinstall" won't even find the old package built from ports. Since making a new version from a new port will overwrite the old files, I tend to just not bother with deinstalling the old version and thus pkg_info becomes a sort of history (the consistency checker can at least tell me which ones are "really" installed).
Ports is nice, I do wish it built more robust packages tho.
FreeBSD has supported >2GB files for as long as I can remember. I don't know how to tell for sure, but it would also appear FreeBSD uses a 3gig user, 1gig kernel address space, since absolute addresses well above 2G were reported available when I ran a malloc arena prospector program.
/compat/linux)
Linux support is superb. I'm posting this reply with linux netscape 4.7, and there's no detectable difference in performance, behavior, or stability. Actually it's more stable than I've ever had it under linux (probably because I'm disinclined to constantly futz with the system installed in
I do have to disappoint you a bit: UFS+softupdates will not give you journalling. Soft updates is merely another block cache and deferred write queue. Speeds up the filesystem tremendously, but doesn't really add any new features.