Getting Darwin to run natively on x86 is no problem. Apple makes an x86 port
available.
The problem is, even if you did manage to emulate well enough to run Quartz, you'd also have to emulate well enough to run all the PPC programs that are the only ones available to use Quartz.
Frankly, it would probably be easier to get GNUstep in sync with the Cocoa api(formerly NeXTStep). Then you could cross-compile Cocoa applications.
It should be possible.
Darwin is open source, under the
APSL, and as you point out, people have modified it to get OS X (even the non-open source parts) to run on pre-G3 machines just fine.
I really wonder how long it will take someone to get OS X running on a non-Apple PPC machine. The code is there, and Darwin is free (as in beer). If you can get Darwin to run on it, Quartz (the closed source part) shouldn't know the difference.
I believe it can be done, and that means that eventually someone will do it.
The book is one of those "based on true events" types.
AFAIK, the book is mostly the story of his life, with some embellishment here and there and presumably some stuff left out to make a better story. The book is upfront about this, and some of the names were changed so as not to embarrass anyone who might come out in a bad light.
One scene in the book that is made up (as I recall) was the one where he saw Sputnik passing over West Virginia. Apparantly it never happened, I think it was cloudy that day, even though it was the catalyst (in the book) for him wanting to build rockets. That scene also gave the movie its title, looking up at the "October Sky" and seeing Sputnik.
As some other people have pointed out, and I will point out here.
The pictures are of the Profile3, and the article is talking about the upcoming Profile4. There are no pictures of the Pofile4 in the article, they just say it will be more iMac-like than the Profile3 (big surprise, a PC company following Apple's lead).
...and I would say that it makes the iPod a minimally functional PDA, but not a full fledged PDA. Everything he (or she) said was true. The complainers here are moaning about the submitter did not point out the limitations of using the iPod as a PDA, but if you've ever picked up an iPod, those limitations are kind of obvious. The iPod is a storage device, and the only input is for navigating what is already there. The iPod is supremely good at that. Asking it to do data entry would be like asking your keyboard to spellcheck.
Actually it is a little more complicated than that.
What the iPod displays when you select a song is the ID3 tags that are stored in the MP3 files. Then this program just creates a bunch of MP3 files with silent sound tracks (and low bitrates) with your contact information stored in the Artist/Album/Title ID3 tags.
I've notice some people complaining that you can't add contacts on the go, so this is worthless. I think that misses the point here. The iPod is an MP3 player. Apple has never claimed that it can do anything else. But if you carry it around with you all the time, this is a nice little hack to add a little extra functionallity. Nobody is claiming that this makes the iPod a full featured PDA.
I just went and downloaded it to give it a whirl. I've been using Mozilla 0.9.8 lately because it gives me fewer problems than IE (IE tends to make sections of pages disappear until I scroll down and then back up).
Chimera is definitely beta, though. I've noticed three bugs already.(I'm using it right now to make this post, so they're not major bugs)
Cursor postion does not update when changed unless you type something.
Pop-up selections do not pop up (you can make selections by selecting a pop-up menu and use the arrow keys)
Preferences menu item is greyed out.
All in all, it is really nice, though. It is already much snappier than Mozilla. I'll be following this one.
This was not an update to Java 1.3.1, it is an update of the MacOS X JVM, which has been running Java 1.3.1 for a while.
Many of the bug fixes here will benefit Apple's implementation of Java 1.4. Apple has been working hard to get their JVM up to snuff and get the hardware Swing acceleration working reliably.
Remember, too, that you cannot de-couple completely the OS and the JVM. This update required that you have the OS X 10.1.3 update already installed.
Someone else said that Apple will have a preview version of 1.4 available at WWDC in May and I'd expect that the production version will probably be part of OS X 10.2 sometime this summer.
I tried following the directions to enable the hardware acceleration on my iBook. I have a program that I wrote that has a table of JTextAreas that I would love to have accelerated.
It didn't seem to work. I'm not sure if the iBook video card is supported. It is not listed in the release notes, but it is listed in the glconfigurationlist.properties file.
If anybody gets it to work, please reply and let everyone know.
It is more than a transistor. The article at the NYTimes (I'm too lazy to link right now), said that IBM had previously anounced a transistor which could switch at 260 GHz and this anouncement is simply the next step, an entire circuit, but probably not a whole CPU.
I think that is correct, but it doesn't mean that someone couldn't make a firewire-native HD. You would just have to put a different controller board on it. It would be more expensive than an IDE drive, just because the controller would have to be "smarter".
I seem to recall that when firewire was being develped (some time before it actually debuted on the B&W powermacs), there was talk about an internal version of firewire that would have a significanly faster transfer rate. Unfortunately, it doesn't ever seem to have come about.
I thought these things were supposed to be easy, and intuitive to ride. I wonder how long the training is.
I'm sure the training involves more than just how to hop on and go. There is probably some maintenance that must be done.
One of the things I remember reading about the Segway company is that they were developing a set of "rules of the road" for their scooters. People will start cursing the scooters if the people who use them just drive like assholes all the time and don't follow any expected rules. They probably will be tring to stress those rules on the early purchasers so the scooter itself does not get a bad reputation.
When I was in high school (10 years ago or so). I went with my father (a CS prof) to some seminars. There we met and talked with some old timers who'd been working with computers since the 50s. They told us about "code libraries" starting back in the days of punch cards. The story went something like this:
Bob: Hey Joe, didn't you write an I/O routine last week.
Joe: Yeah. [Joe pulls a stack of punch cards down off a shelf with a rubber band around them, and hands them to Bob.]
Bob: Thanks. [Bob removes the rubber band and inserts the stack of punch cards into the program he is writing.]
I think that the general idea here is that, though you can say anything you want, you can be held accountable for what you say.
For example: Threatening to kill someone is illegal. It is an act of speech, and you have every right to make it, but you can be held responsible for it.
If you say anything with the intent to cause harm to an individual or a group of people by saying it, you can be held accountable. All the things I cited fall under that category.
I agree with you, but my point was that if, 8 months from now one of your IDE controllers dies and you cannot get the exact same one to replace it (for whatever reason), swapping in a new one *could* cause problems, and reconfiguring the kernel for that machine *could* fsck up the cluster.
I'm not saying you can't build a good Linux cluster, I'm saying that maintenence on a Linux cluster is more complicated than maintenence on a Mac cluster. Can we agree on that?
Besides some people would pay to have someone else sweat the details for them. That's why Apple is still in buisness. How much would you charge to set up a Linux cluster for someone else? How much would that cut into the price differential between LinuxPCs and MacOS X boxes?
I wasn't claming that the MacOS kernel would do such a thing. (Though the Mach kernel at the heart of MacOS X runs on a lot of architectures)
The original comparison here was maintenance on commodity PC hardware vs a Mac. OS support of MacOS on a Mac is assured (at least for the useful life of a cluster). You can't say the same for Linux on just any old PC hardware you slap together.
My statement was that it is not as much of a sure thing getting Linux running properly on a frankenPC as it is getting MacOS configured on a Mac.
On a Mac, if the OS doesn't get along with the default hardware, you only have one source to complain to: Apple, and they have to fix whatever problem you have if the machine is under warrenty.
I was just pointing out that there are limits to free speech under the US Constitution.
Anyway, I'm just recalling what I learned in my high school government class in 1994. The four examples I gave are all types of speech that the Supreme Court has issued rulings saying that they are not protected under the 1st amendment. Whether you disagree with the court on those points is a different debate.
The constitution is interpreted by the Supreme Court, and it is not as absolute as that.
There are several types of speech which are not protected including:
Slander
Libel
Hate Speech
Fighting Words
Any of these can be, and are abridged by law. Do you think someone who doesn't like you should be able to go around telling lies that get you fired? (for example)
Care to suggest a "stable" kernel that will run on any machine you throw at it?
No matter what NIC chipset? Any video card? All I/O configureations? If you put it into a cluster and there were network problems could you guarantee that it was a bad NIC and not a misconfigured kernel?
I know you can buy a PC with a warranty, but the vast majority of PC vendors do not support the OS themselves. So you can quickly get into a situation where the hardware people will tell you it is a software problem and the software people will tell you it is a hardware problem. You can end up SOL that way. Apple can't do that to you.
Glad to know you managed to put together a computer that did not have any problems. I have an 7.5 year old PowerMac that had the power supply replaced after 3 years but still runs just fine. Doesn't prove anything.
I'll agree with you on a few of your points, but not this one:
-- Maintenance costs are much, much lower. Anything goes wrong with a PC node, just swap out that part with another commodity part. Mac repair or parts replacement costs will eat you, especially if you start to have many, many nodes.
Here's the thing. If you buy a Mac (any Mac), to run any of the MacOSes (classic or X), you expect the thing to work, and work flawlessly. You do not have to worry about driver conflicts, you do not have to worry if the Kernel version works correctly with your hardware, and you really should not expect any compoent failures, given the price you are paying.
Plus, every Mac comes with a 1 year warrenty. The whole thing is guaranteed by Apple to Just Work for a whole year. If one of the nodes doesn't work right, you send it back to Apple and they either fix it, or send you a new one.
Besides, if you are doing clustering computing, by the time the Macs start failing, you'll be putting together a new cluster anyway. By the time that starts happening, you can just sell the old cluster on the used market, machine by machine, and probably build a cluster 5x as powerful at a similar price to what you paid for the last one.
Stan Williams, one of the guys mentioned here, came to my department to give a talk about the work they were doing at HP about 2 years ago.
What they are doing is really facinating, and it's not quite as simple as just re-programming the chip when they come off the line. The chips will continue to develop defects, even during service.
The way they get around this is to design a fault tolerant processing scheme. When you drop the sizes down as much as these people are, you get a several order of magnitude increase in the number of transistors, so you can afford to have the chip do the same calculation, say 500 times in different sections of the chip. The chip itself can figure out what sections are bad, and stop using them on its own.
HP actually built a full size computer where they designed some ASICs that computed using lookup tables (!). They had them fabed and asked the fab to send them the defective chips along with the good ones. They then mixed the good chips and bad chips together (I think it was like a 1/2 good/bad ratio) and hired a high school student to hook up the wiring. Now keep in mind that even on the "defective" chips, part of the chip still worked. It only takes 1 defect to spoil a traditional chip. On the whole, the components on the chips had about a 3% defect rate.
The whole thing ran at a whopping 1MHz and may not have been wired up exactly to specifications, but it was "programmed" with a standard computer first to find the defects and route around them. Performance wise, it was on par with the fastest HP workstations of the day. (there's the MHz myth for you)
So the idea here is to design chips that have so many circuits that you can afford to build in fault tolerance. What is more, you can afford to have the chips constantly checking themselves looking for new faults.
In short, zero defect tolerence is not necessarily a good thing. One defect in one transistor can render a Pentium processor worthless. The smaller you make them, and the more transistors you add, the harder it will be to achieve defect free parts. Yields go down, price goes up.
And if you don't believe me, they published an article in Science about the computer they built (it was called Teramac IIRC)
The PTO is rewarded for processing as many patents as possible, the problem is just that it is much faster to approve a patent application than it is to reject it, if only because if they reject it then they have to deal with appeals. If they approve it, then anyone contesting the patent must challenge it in court, not at the PTO.
You might not have to screw in a VGA connector on a desktop, but if you have your laptop connected to a projector for a presentation, you definitely need it screwed in or it will pull right out at the slightest tug. The standard VGA connector was not designed to hold itself tightly in place. That is why there are screws on it.
Anyway, I've spent plenty of time trying to plug VGA connectors in when I couldn't see the port and found I was trying to plug it in backwards. VGA is only easy to attach if you can see the port. USB is the same way, you can easily look the port and figure out which way the plug goes in, but if you can't see it, it can be tough. Not to mention how hard a VGA connector is to plug in if you bend one of the pins. Thats a problem you'll never see on USB, firewire, or the Apple miniVGA port.
Firewire is the champ here, BTW. It has a connector that is easy to figure out and holds in place quite well. The miniVGA port on the iBook is very similar, the plug is chamfered on one side and it makes it easy to look at it and the plug and figure out which way it goes.
I actually own an iBook (a 12" DVD model), and I can tell you that there is not room for a full size VGA port on this thing. All the available space is taken up by the DVD, battery, HD, the screen hinge, or the ports (modem/ethernet/firewire/usb/usb/miniVGA/headphone -video). If you switched the miniVGA port to a full size port, then the ports would intrude into the space where the HD is.
Besides which, screwing a VGA connector on is a PITA on a laptop. With this design, you can screw the adapter onto the monitor cable and then just plug it in. Apple provides the adaptor and it is as easy to plug and unplug as a USB or Firewire cable.
Slashdot Mirror
The problem is, even if you did manage to emulate well enough to run Quartz, you'd also have to emulate well enough to run all the PPC programs that are the only ones available to use Quartz.
Frankly, it would probably be easier to get GNUstep in sync with the Cocoa api(formerly NeXTStep). Then you could cross-compile Cocoa applications.
I really wonder how long it will take someone to get OS X running on a non-Apple PPC machine. The code is there, and Darwin is free (as in beer). If you can get Darwin to run on it, Quartz (the closed source part) shouldn't know the difference.
I believe it can be done, and that means that eventually someone will do it.
AFAIK, the book is mostly the story of his life, with some embellishment here and there and presumably some stuff left out to make a better story. The book is upfront about this, and some of the names were changed so as not to embarrass anyone who might come out in a bad light.
One scene in the book that is made up (as I recall) was the one where he saw Sputnik passing over West Virginia. Apparantly it never happened, I think it was cloudy that day, even though it was the catalyst (in the book) for him wanting to build rockets. That scene also gave the movie its title, looking up at the "October Sky" and seeing Sputnik.
Incidently, "October Sky" is an anagram for "Rocket Boys".
The pictures are of the Profile3, and the article is talking about the upcoming Profile4. There are no pictures of the Pofile4 in the article, they just say it will be more iMac-like than the Profile3 (big surprise, a PC company following Apple's lead).
...and I would say that it makes the iPod a minimally functional PDA, but not a full fledged PDA. Everything he (or she) said was true. The complainers here are moaning about the submitter did not point out the limitations of using the iPod as a PDA, but if you've ever picked up an iPod, those limitations are kind of obvious. The iPod is a storage device, and the only input is for navigating what is already there. The iPod is supremely good at that. Asking it to do data entry would be like asking your keyboard to spellcheck.
What the iPod displays when you select a song is the ID3 tags that are stored in the MP3 files. Then this program just creates a bunch of MP3 files with silent sound tracks (and low bitrates) with your contact information stored in the Artist/Album/Title ID3 tags.
I've notice some people complaining that you can't add contacts on the go, so this is worthless. I think that misses the point here. The iPod is an MP3 player. Apple has never claimed that it can do anything else. But if you carry it around with you all the time, this is a nice little hack to add a little extra functionallity. Nobody is claiming that this makes the iPod a full featured PDA.
Chimera is definitely beta, though. I've noticed three bugs already.(I'm using it right now to make this post, so they're not major bugs)
All in all, it is really nice, though. It is already much snappier than Mozilla. I'll be following this one.
This was not an update to Java 1.3.1, it is an update of the MacOS X JVM, which has been running Java 1.3.1 for a while.
Many of the bug fixes here will benefit Apple's implementation of Java 1.4. Apple has been working hard to get their JVM up to snuff and get the hardware Swing acceleration working reliably.
Remember, too, that you cannot de-couple completely the OS and the JVM. This update required that you have the OS X 10.1.3 update already installed.
Someone else said that Apple will have a preview version of 1.4 available at WWDC in May and I'd expect that the production version will probably be part of OS X 10.2 sometime this summer.
It didn't seem to work. I'm not sure if the iBook video card is supported. It is not listed in the release notes, but it is listed in the glconfigurationlist.properties file.
If anybody gets it to work, please reply and let everyone know.
It is more than a transistor. The article at the NYTimes (I'm too lazy to link right now), said that IBM had previously anounced a transistor which could switch at 260 GHz and this anouncement is simply the next step, an entire circuit, but probably not a whole CPU.
I seem to recall that when firewire was being develped (some time before it actually debuted on the B&W powermacs), there was talk about an internal version of firewire that would have a significanly faster transfer rate. Unfortunately, it doesn't ever seem to have come about.
I'm sure the training involves more than just how to hop on and go. There is probably some maintenance that must be done.
One of the things I remember reading about the Segway company is that they were developing a set of "rules of the road" for their scooters. People will start cursing the scooters if the people who use them just drive like assholes all the time and don't follow any expected rules. They probably will be tring to stress those rules on the early purchasers so the scooter itself does not get a bad reputation.
For example: Threatening to kill someone is illegal. It is an act of speech, and you have every right to make it, but you can be held responsible for it.
If you say anything with the intent to cause harm to an individual or a group of people by saying it, you can be held accountable. All the things I cited fall under that category.
I'm not saying you can't build a good Linux cluster, I'm saying that maintenence on a Linux cluster is more complicated than maintenence on a Mac cluster. Can we agree on that?
Besides some people would pay to have someone else sweat the details for them. That's why Apple is still in buisness. How much would you charge to set up a Linux cluster for someone else? How much would that cut into the price differential between LinuxPCs and MacOS X boxes?
I wasn't claming that the MacOS kernel would do such a thing. (Though the Mach kernel at the heart of MacOS X runs on a lot of architectures)
The original comparison here was maintenance on commodity PC hardware vs a Mac. OS support of MacOS on a Mac is assured (at least for the useful life of a cluster). You can't say the same for Linux on just any old PC hardware you slap together.
My statement was that it is not as much of a sure thing getting Linux running properly on a frankenPC as it is getting MacOS configured on a Mac.
On a Mac, if the OS doesn't get along with the default hardware, you only have one source to complain to: Apple, and they have to fix whatever problem you have if the machine is under warrenty.
Anyway, I'm just recalling what I learned in my high school government class in 1994. The four examples I gave are all types of speech that the Supreme Court has issued rulings saying that they are not protected under the 1st amendment. Whether you disagree with the court on those points is a different debate.
There are several types of speech which are not protected including:
Any of these can be, and are abridged by law. Do you think someone who doesn't like you should be able to go around telling lies that get you fired? (for example)
No matter what NIC chipset? Any video card? All I/O configureations? If you put it into a cluster and there were network problems could you guarantee that it was a bad NIC and not a misconfigured kernel?
I know you can buy a PC with a warranty, but the vast majority of PC vendors do not support the OS themselves. So you can quickly get into a situation where the hardware people will tell you it is a software problem and the software people will tell you it is a hardware problem. You can end up SOL that way. Apple can't do that to you.
Glad to know you managed to put together a computer that did not have any problems. I have an 7.5 year old PowerMac that had the power supply replaced after 3 years but still runs just fine. Doesn't prove anything.
Here's the thing. If you buy a Mac (any Mac), to run any of the MacOSes (classic or X), you expect the thing to work, and work flawlessly. You do not have to worry about driver conflicts, you do not have to worry if the Kernel version works correctly with your hardware, and you really should not expect any compoent failures, given the price you are paying.
Plus, every Mac comes with a 1 year warrenty. The whole thing is guaranteed by Apple to Just Work for a whole year. If one of the nodes doesn't work right, you send it back to Apple and they either fix it, or send you a new one.
Besides, if you are doing clustering computing, by the time the Macs start failing, you'll be putting together a new cluster anyway. By the time that starts happening, you can just sell the old cluster on the used market, machine by machine, and probably build a cluster 5x as powerful at a similar price to what you paid for the last one.
What they are doing is really facinating, and it's not quite as simple as just re-programming the chip when they come off the line. The chips will continue to develop defects, even during service.
The way they get around this is to design a fault tolerant processing scheme. When you drop the sizes down as much as these people are, you get a several order of magnitude increase in the number of transistors, so you can afford to have the chip do the same calculation, say 500 times in different sections of the chip. The chip itself can figure out what sections are bad, and stop using them on its own.
HP actually built a full size computer where they designed some ASICs that computed using lookup tables (!). They had them fabed and asked the fab to send them the defective chips along with the good ones. They then mixed the good chips and bad chips together (I think it was like a 1/2 good/bad ratio) and hired a high school student to hook up the wiring. Now keep in mind that even on the "defective" chips, part of the chip still worked. It only takes 1 defect to spoil a traditional chip. On the whole, the components on the chips had about a 3% defect rate.
The whole thing ran at a whopping 1MHz and may not have been wired up exactly to specifications, but it was "programmed" with a standard computer first to find the defects and route around them. Performance wise, it was on par with the fastest HP workstations of the day. (there's the MHz myth for you)
So the idea here is to design chips that have so many circuits that you can afford to build in fault tolerance. What is more, you can afford to have the chips constantly checking themselves looking for new faults.
In short, zero defect tolerence is not necessarily a good thing. One defect in one transistor can render a Pentium processor worthless. The smaller you make them, and the more transistors you add, the harder it will be to achieve defect free parts. Yields go down, price goes up.
And if you don't believe me, they published an article in Science about the computer they built (it was called Teramac IIRC)
The PTO is rewarded for processing as many patents as possible, the problem is just that it is much faster to approve a patent application than it is to reject it, if only because if they reject it then they have to deal with appeals. If they approve it, then anyone contesting the patent must challenge it in court, not at the PTO.
Anyway, I've spent plenty of time trying to plug VGA connectors in when I couldn't see the port and found I was trying to plug it in backwards. VGA is only easy to attach if you can see the port. USB is the same way, you can easily look the port and figure out which way the plug goes in, but if you can't see it, it can be tough. Not to mention how hard a VGA connector is to plug in if you bend one of the pins. Thats a problem you'll never see on USB, firewire, or the Apple miniVGA port.
Firewire is the champ here, BTW. It has a connector that is easy to figure out and holds in place quite well. The miniVGA port on the iBook is very similar, the plug is chamfered on one side and it makes it easy to look at it and the plug and figure out which way it goes.
Besides which, screwing a VGA connector on is a PITA on a laptop. With this design, you can screw the adapter onto the monitor cable and then just plug it in. Apple provides the adaptor and it is as easy to plug and unplug as a USB or Firewire cable.