Just to be clear, I'm grouping the Quartz/PDF technology in with the other technologies on the same layer, which no longer jives with Apple's official architecture terminology, which you can see in all its gumdrop glory here.
The real problem porting would be Quartz. From what I understand, Quartz is rather heavily optimized for AltiVec.
Quartz is partially rooted in OpenStep's Display PostScript, which already runs on Intel. The main differences between Quartz and Display PostScript are elements supporting PDF, QuickTime, OpenGL and QuickDraw. Of those, only QuickDraw is unavailable on Intel in a previous form. Quartz also takes advantage of nVidia and ATI cards. Thus, most of Quartz already exists for the x86 platform, so it is more likely to be among the more trivial parts of the port.
AltiVec optimization is available for much of Mac OS X resulting is speed enhancements, but no part of Mac OS X is dependent upon AltiVec. Mac OS X runs fine on G3s without AltiVec, which pretty much disproves your assertion.
Hardware may be a "commoddity," but even with Apple's "fan club" niche, it is still one of the world's top computer makers. Apple may only have 9% of the desktop market, but Dell only has 10%. In the hardware industry, this means whatever Apple is doing is working splendidly. Plus, Dell has to pay a good chunk of its profits to Microsoft, whereas Apple collects residual income from OS upgrades and pays nothing to Microsoft beyond dividends.
What is your definition of "marginal finances?" Even in Apple's darkest days last millennium, it still had $9 billion in cash, which was only down to $8 billion-ish by the time the original iMac brought Apple back into profitability. Apple was burning through $250 million a quarter, but could have easily survived that for 28 years before getting into serious financial trouble. After the tech bubble busted, Apple has emerged financially as one of the strongest companies in the technology sector.
Perhaps you are still looking exclusively at market caps, which is the mentality that created the bubble economy and its subsequent bust. Market capitalization is only one of many factors, and probably the most deceptive, especially when evaluating Apple whose stock prices are almost always artificially depressed due to uninformed prejudices surrounding its products.
Please don't think I'm just picking on you-- technical savants tend to be ignorant of the most elementary business principles, which is why they were suckers for the mythical "New Economy." The non-existant "New Economy" was nothing more than an illusion created for the sole purpose of separating fools from their money. Those of us who knew that the emperor had no clothes made a killing off the plummeting stock market while the rest of you lost your shirts.
The irrationally exuberant have no one but themselves to blame!
Okay, this is old territory. Mac OS X is based on a PowerPC port of OpenStep, which was an Intel port of NeXTstep-- so at first blush one would think the transition simple. However, there are portions of Mac OS X which rely heavily on the PowerPC hardware, primarily the Classic and Carbon environments. Classic and Carbon could either be eliminated or emulated on Intel.
Eliminating Classic and Carbon would mean that the main work would be in upgrading OpenStep to the Mac additions to Cocoa, Aqua, etc., so we could call any theoretic Intel port of Mac OS X "OpenStep X". OpenStep X would be able to compile the source code for all the best new Mac software, but it would not be backward-compatible with Classic or Carbonized apps, which accounts for the vast majority of existing Mac software.
The alternative, emulating Carbon and Classic on Intel, would result, at best, in performance approximately similar to running Windows XP on Virtual PC on a Mac, which is horrendously slow (I do it on my PowerBook all too often).
The main reason Apple won't make OpenStep X, beyond the fact that it knows hardware is its core business and can only be hurt by such a move, is that the old OpenStep already runs much faster on Intel chips than OS X on the latest G4's, so OpenStep X could easily undercut Apple's core business (Porche metaphors notwithstanding). Emulating Classic and Carbon would make OpenStep slower than OS X, but would require a mammoth effort and result in nothing but bad press for Apple.
As much as Intel users might salivate over Mac OS X, there is only one realistic way to use Mac OS X, and that is to buy Mac hardware. iMacs and iBooks can hardly be accused of being overpriced, nor can they be accused of being overly expandable. PowerMacs and PowerBooks provide a pretty good value considering all that's included. Macs use regular USB, FireWire and VGA ports as well as PC Card or PCI slots compatible with a majority of Windows cards.
Now that I've basically poo-pooed the idea, I believe there is one way Apple could throw a bone to the Intel dogs without having to pay their vetrinary bills-- release the old OpenStep (no Carbon, Classic, Aqua or Mac compatibility) source code under the GNU or Darwin license to get geeks salivating over it. No matter what we do with OpenStep, everyone will probably come to the conclusion that it is great but Mac OS X is far better and worth the price tag of Mac hardware. However, that is a risk with the potential to backfire, maybe as an OpenStep-Linux hybrid mounting a more serious challenge to Windows on PCs and cutting into sales of Macs. I doubt Apple will take such a risk, but Steve Jobs is the same guy who moved NeXTstep to OpenStep, so who can say?
Back in 1979-80 I noticed (through perusing manuals) that my TRS-80 Model III had a programmable oscillator to help generate analog signals for peripheral interfaces, including the cassette interface. My first project was to program the oscillator to generate music and sound effects over AM and FM frequencies. That was fun for awhile, but then I figured out how to send just about any sound wave I wanted over the cassette drive cable, so if I depressed the copy protection lever with my finger while I pressed record and play, I could use my boom box for better sound. I wrote up my findings as a science fair project, got all kinds of press covereage, and the next thing you know, everybody was doing it.
My next science fair project was building a multiplatform (Apple II, TI 99/4, Vic20, Atari 400, TRS-80 models III and II) network over cassette drive cables and RS-232 serial ports for the purpose of playing games, sharing and printing formatted documents with hyperlinks, and accessing incompatible peripherals. I narrowly missed advancing to the international science fair on that one, probably because it was ahead of its time.
The judges two decades ago didn't like to admit they knew nothing about computers-- the comments I got were along the lines of "the ability to use computers is a good skill but does not constitute science unless it presents something new or investigates something less trivial," writing off my creative ideas and all the code I wrote as just the way people use computers. A decade later, the same idea was no longer called "trivial," but "the single most important technological innovation of our generation."
This paper is a great germ for further study, but until there is some practical experimentation to support it, it's hot air. Good hot air, perhaps, but still vapor. If theoreticians want the support of engineering types, we need to see some experimental tests. I've seen dozens of really good theories on improving peer and grid networks that turn sour as soon as anyone tries to implement them. This paper mentions much about peer networks and nothing about peer review. I'm withholding judgement until I read the peer review.
Yeah, I still tote an HP48GX and a Newton because they're easy to program (for me) and can use CF cards and PCMCIA cards. The old TRS-80s harken back to a day when the only user interface was BASIC, so if you needed any software or drivers, you just rolled your own-- back when the OS ran comfortably from 32k ROMs and 16k-48k of RAM was all you needed to do just about any application (witness visicalc, superscripsit, dbase) without hires graphics. IMHO, as much sophistication as we have gained in the intervening years, the changes are not always improvements. I'd trash my scicalc and PDA if I could get my hands on the (ahem) power/flexibility of a TRS-80 Model 100 in a device like the old REX (PCMCIA-sized PDA). Know of any such animal?
Candy, candies. Sandy, sandies. Tandy, tandies. Tomaeto, Tomahto, Potaeto, Potahtoe (Quaylism). Let's call the whole thing off. This is the internet, IMHO. Since when do we care about grammour?
Re:Ah, the Model 100
on
Tandys Never Die
·
· Score: 2, Informative
The original batch of Model 100's were under the TRS-80 brand. Then, Tandy Corp. dropped the TRS-80 brand from its entire line, when it became the Tandy Model 100. Then, the word "Model" was removed, and became the Tandy 100, then the Tandy 102. Clamshell versions were released under the Tandy 2xx line. As mentioned earlier, these were closer to souped-up TRS-80 Pocket Computers rather than scaled-down TRS-80 desktops. Meaning, Tandy 100's didn't run TRS-DOS. This doesn't mean much because TRS-DOS was customized for each model leaving 5 relatively incompatible families of TRS-80 machines: Models I, III and the IV/4 lines; Models II, 12, 16, etc.; Color Computers; Pocket Computers; and Models 10x-2xx.
(from a RadioShack alum and Tandy enthusiast... won my first few amateur programming competitions on Trash-80s)
The method I use is CGI graphics gatekeeper (checking referer before sending graphic). Same basic algorithm as FormMail's to block all except specific referers. The graphic remains in cache so long as the URL ends with the right extension. For example, "http://cnn.com/cgi-bin/ggk.cgi?shortpath/to/graph ic.gif" stays in cache, but "http://cnn.com/cgi-bin/ggk.cgi?g=graphic.gif&some thing=value" does not.
If your site has high traffic or if you just want to get fancy and more efficient, code or license a plugin like Akamai's or Angelfire's. Theirs send an ad graphic if anyone attempts to remotely source an image from their servers. To test, view a Google cache of an Angelfire page.
Right. The english word "link" is heavily overloaded. We geeks need to come up with different words for hyperlinking with an HREF attribute vs. linking an image (or stylesheet or other remote file) inline with a SRC attribute. Maybe the attributes themselves hold the answer, "referencing" vs. "sourcing." Referencing is fair game, but remote sourcing requires permission.
Also, the analogies supplied here all fall a little short. It seems that telephony provides a better model... toll-free numbers (sourcing) vs. long distance (referencing). Whenever you dial a toll-free number, even if you have caller ID blocking, your phone number shows up on the remote caller ID giving the toll-free service an opportunity not to accept charges associated with your call. On the other hand, with long distance, the caller is footing the bill, so the caller can block their number from showing up on the remote caller ID boxes.
In case the analogy isn't obvious, it should be the responsibility of the web server to block remote sourcing if they don't like it, otherwise remote sourcing is fair game. Similarly, it should be the web designer's responsibility to design pages that break framesets if they don't want their pages framed. The solution to the problem should be handled with available technology, not an unconstitution court ruling restricting free expression.
I have a feeling that this is the last chance for Amiga, it is sink or swim. If they don't succeed this time, then it is all over for the platform.
If I had one yen for every time I've heard that pronouncement, I'd still be a poor man, but I'd have a million yen. The poster who said "it just won't die" was more realistic. The Amiga inspired so many geeks that we'll keep trying to revive it just to play with it. I'm pretty sure the goal of Amiga-wraiths is no longer to bring down Microsoft, but nostalgia.
After my Amigas became doorstops, I moved to Macs. Then everybody began pronouncing the imminent death of Apple, with each new technology being a make-or-break litmus test. "If OpenDoc doesn't work out, Apple is dead." "If PCs don't adopt RISC, Apple is dead." "If CHReP doesn't take off, Apple's dead." "Unless the clones succeed, it's curtains for Apples." "If the 20th Anniversary Mac tanks, so does Apple." "If the Newton doesn't survive, neither will Apple." "If Apple can't leverage HyperCard as an HTML platform, chapter 11 is around the corner."
All those technologies failed miserably in the marketplace-- yet, Apple is still around and doing quite nicely in its niche market. Commodore bit the big one ages ago (in web weeks), yet the Amiga still has a dedicated hardcore cult following. Harley-Davidson bikes are unreliable compared to rice rockets, yet there is still a subculture centered around them. Amiga enthusiasts are mostly in our 30s and 40s, so the AmigaOS still has about 40 years to get its act together before we all croak.
Unfortunately they chose QNX as their kernel, which is not only proprietary, but also has few fanatical supporters. (unlike either *BSD or Linux, both of which have lots of fanatical supporters.)
The Amiga has plenty of its own fanatical supporters, thank-you-very-much. Sure, Amiga on Linux would be the ultimate geek toy, but "BSD fanatic" is an oxymoron compared to "Amiga fanatic." We're the people who got our kicks going to Apple and PC stores just to watch the salespeople squirm when we started comparing them to Amigas.
What if MS were to apply the Apple Mac OS X model to Windows, using GNU Linux instead of BSD? Darwin is (nominally) open, yet Apple retains control through its Aqua/Carbon/Cocoa model.
Let's call a theoretical GNU Linux layer "Dante" (similar to Darwin), the MS-Windows backward-compatibility emulation environment "Wintel" (like Classic) and the new part of the OS "Winux" (like Carbon/Cocoa/Aqua/X-Windows).
Couldn't such an OS allow MS to retain its monopoly while effectively seizing control of Linux? Would this be a victory for Linux, or would Dante destroy everything penguinheads hold dear?
Slashdot Mirror
Just to be clear, I'm grouping the Quartz/PDF technology in with the other technologies on the same layer, which no longer jives with Apple's official architecture terminology, which you can see in all its gumdrop glory here.
The real problem porting would be Quartz. From what I understand, Quartz is rather heavily optimized for AltiVec.
Quartz is partially rooted in OpenStep's Display PostScript, which already runs on Intel. The main differences between Quartz and Display PostScript are elements supporting PDF, QuickTime, OpenGL and QuickDraw. Of those, only QuickDraw is unavailable on Intel in a previous form. Quartz also takes advantage of nVidia and ATI cards. Thus, most of Quartz already exists for the x86 platform, so it is more likely to be among the more trivial parts of the port.
AltiVec optimization is available for much of Mac OS X resulting is speed enhancements, but no part of Mac OS X is dependent upon AltiVec. Mac OS X runs fine on G3s without AltiVec, which pretty much disproves your assertion.
Hardware may be a "commoddity," but even with Apple's "fan club" niche, it is still one of the world's top computer makers. Apple may only have 9% of the desktop market, but Dell only has 10%. In the hardware industry, this means whatever Apple is doing is working splendidly. Plus, Dell has to pay a good chunk of its profits to Microsoft, whereas Apple collects residual income from OS upgrades and pays nothing to Microsoft beyond dividends.
What is your definition of "marginal finances?" Even in Apple's darkest days last millennium, it still had $9 billion in cash, which was only down to $8 billion-ish by the time the original iMac brought Apple back into profitability. Apple was burning through $250 million a quarter, but could have easily survived that for 28 years before getting into serious financial trouble. After the tech bubble busted, Apple has emerged financially as one of the strongest companies in the technology sector.
Perhaps you are still looking exclusively at market caps, which is the mentality that created the bubble economy and its subsequent bust. Market capitalization is only one of many factors, and probably the most deceptive, especially when evaluating Apple whose stock prices are almost always artificially depressed due to uninformed prejudices surrounding its products.
Please don't think I'm just picking on you-- technical savants tend to be ignorant of the most elementary business principles, which is why they were suckers for the mythical "New Economy." The non-existant "New Economy" was nothing more than an illusion created for the sole purpose of separating fools from their money. Those of us who knew that the emperor had no clothes made a killing off the plummeting stock market while the rest of you lost your shirts.
The irrationally exuberant have no one but themselves to blame!
Okay, this is old territory. Mac OS X is based on a PowerPC port of OpenStep, which was an Intel port of NeXTstep-- so at first blush one would think the transition simple. However, there are portions of Mac OS X which rely heavily on the PowerPC hardware, primarily the Classic and Carbon environments. Classic and Carbon could either be eliminated or emulated on Intel.
Eliminating Classic and Carbon would mean that the main work would be in upgrading OpenStep to the Mac additions to Cocoa, Aqua, etc., so we could call any theoretic Intel port of Mac OS X "OpenStep X". OpenStep X would be able to compile the source code for all the best new Mac software, but it would not be backward-compatible with Classic or Carbonized apps, which accounts for the vast majority of existing Mac software.
The alternative, emulating Carbon and Classic on Intel, would result, at best, in performance approximately similar to running Windows XP on Virtual PC on a Mac, which is horrendously slow (I do it on my PowerBook all too often).
The main reason Apple won't make OpenStep X, beyond the fact that it knows hardware is its core business and can only be hurt by such a move, is that the old OpenStep already runs much faster on Intel chips than OS X on the latest G4's, so OpenStep X could easily undercut Apple's core business (Porche metaphors notwithstanding). Emulating Classic and Carbon would make OpenStep slower than OS X, but would require a mammoth effort and result in nothing but bad press for Apple.
As much as Intel users might salivate over Mac OS X, there is only one realistic way to use Mac OS X, and that is to buy Mac hardware. iMacs and iBooks can hardly be accused of being overpriced, nor can they be accused of being overly expandable. PowerMacs and PowerBooks provide a pretty good value considering all that's included. Macs use regular USB, FireWire and VGA ports as well as PC Card or PCI slots compatible with a majority of Windows cards.
Now that I've basically poo-pooed the idea, I believe there is one way Apple could throw a bone to the Intel dogs without having to pay their vetrinary bills-- release the old OpenStep (no Carbon, Classic, Aqua or Mac compatibility) source code under the GNU or Darwin license to get geeks salivating over it. No matter what we do with OpenStep, everyone will probably come to the conclusion that it is great but Mac OS X is far better and worth the price tag of Mac hardware. However, that is a risk with the potential to backfire, maybe as an OpenStep-Linux hybrid mounting a more serious challenge to Windows on PCs and cutting into sales of Macs. I doubt Apple will take such a risk, but Steve Jobs is the same guy who moved NeXTstep to OpenStep, so who can say?
My next science fair project was building a multiplatform (Apple II, TI 99/4, Vic20, Atari 400, TRS-80 models III and II) network over cassette drive cables and RS-232 serial ports for the purpose of playing games, sharing and printing formatted documents with hyperlinks, and accessing incompatible peripherals. I narrowly missed advancing to the international science fair on that one, probably because it was ahead of its time.
The judges two decades ago didn't like to admit they knew nothing about computers-- the comments I got were along the lines of "the ability to use computers is a good skill but does not constitute science unless it presents something new or investigates something less trivial," writing off my creative ideas and all the code I wrote as just the way people use computers. A decade later, the same idea was no longer called "trivial," but "the single most important technological innovation of our generation."
Go figure.
This paper is a great germ for further study, but until there is some practical experimentation to support it, it's hot air. Good hot air, perhaps, but still vapor. If theoreticians want the support of engineering types, we need to see some experimental tests. I've seen dozens of really good theories on improving peer and grid networks that turn sour as soon as anyone tries to implement them. This paper mentions much about peer networks and nothing about peer review. I'm withholding judgement until I read the peer review.
Yeah, I still tote an HP48GX and a Newton because they're easy to program (for me) and can use CF cards and PCMCIA cards. The old TRS-80s harken back to a day when the only user interface was BASIC, so if you needed any software or drivers, you just rolled your own-- back when the OS ran comfortably from 32k ROMs and 16k-48k of RAM was all you needed to do just about any application (witness visicalc, superscripsit, dbase) without hires graphics. IMHO, as much sophistication as we have gained in the intervening years, the changes are not always improvements. I'd trash my scicalc and PDA if I could get my hands on the (ahem) power/flexibility of a TRS-80 Model 100 in a device like the old REX (PCMCIA-sized PDA). Know of any such animal?
Candy, candies. Sandy, sandies. Tandy, tandies. Tomaeto, Tomahto, Potaeto, Potahtoe (Quaylism). Let's call the whole thing off. This is the internet, IMHO. Since when do we care about grammour?
The original batch of Model 100's were under the TRS-80 brand. Then, Tandy Corp. dropped the TRS-80 brand from its entire line, when it became the Tandy Model 100. Then, the word "Model" was removed, and became the Tandy 100, then the Tandy 102. Clamshell versions were released under the Tandy 2xx line. As mentioned earlier, these were closer to souped-up TRS-80 Pocket Computers rather than scaled-down TRS-80 desktops. Meaning, Tandy 100's didn't run TRS-DOS. This doesn't mean much because TRS-DOS was customized for each model leaving 5 relatively incompatible families of TRS-80 machines: Models I, III and the IV/4 lines; Models II, 12, 16, etc.; Color Computers; Pocket Computers; and Models 10x-2xx.
(from a RadioShack alum and Tandy enthusiast... won my first few amateur programming competitions on Trash-80s)The method I use is CGI graphics gatekeeper (checking referer before sending graphic). Same basic algorithm as FormMail's to block all except specific referers. The graphic remains in cache so long as the URL ends with the right extension. For example, "http://cnn.com/cgi-bin/ggk.cgi?shortpath/to/graph ic.gif" stays in cache, but "http://cnn.com/cgi-bin/ggk.cgi?g=graphic.gif&some thing=value" does not.
If your site has high traffic or if you just want to get fancy and more efficient, code or license a plugin like Akamai's or Angelfire's. Theirs send an ad graphic if anyone attempts to remotely source an image from their servers. To test, view a Google cache of an Angelfire page.
Right. The english word "link" is heavily overloaded. We geeks need to come up with different words for hyperlinking with an HREF attribute vs. linking an image (or stylesheet or other remote file) inline with a SRC attribute. Maybe the attributes themselves hold the answer, "referencing" vs. "sourcing." Referencing is fair game, but remote sourcing requires permission.
Also, the analogies supplied here all fall a little short. It seems that telephony provides a better model... toll-free numbers (sourcing) vs. long distance (referencing). Whenever you dial a toll-free number, even if you have caller ID blocking, your phone number shows up on the remote caller ID giving the toll-free service an opportunity not to accept charges associated with your call. On the other hand, with long distance, the caller is footing the bill, so the caller can block their number from showing up on the remote caller ID boxes.
In case the analogy isn't obvious, it should be the responsibility of the web server to block remote sourcing if they don't like it, otherwise remote sourcing is fair game. Similarly, it should be the web designer's responsibility to design pages that break framesets if they don't want their pages framed. The solution to the problem should be handled with available technology, not an unconstitution court ruling restricting free expression.
If I had one yen for every time I've heard that pronouncement, I'd still be a poor man, but I'd have a million yen. The poster who said "it just won't die" was more realistic. The Amiga inspired so many geeks that we'll keep trying to revive it just to play with it. I'm pretty sure the goal of Amiga-wraiths is no longer to bring down Microsoft, but nostalgia.
After my Amigas became doorstops, I moved to Macs. Then everybody began pronouncing the imminent death of Apple, with each new technology being a make-or-break litmus test. "If OpenDoc doesn't work out, Apple is dead." "If PCs don't adopt RISC, Apple is dead." "If CHReP doesn't take off, Apple's dead." "Unless the clones succeed, it's curtains for Apples." "If the 20th Anniversary Mac tanks, so does Apple." "If the Newton doesn't survive, neither will Apple." "If Apple can't leverage HyperCard as an HTML platform, chapter 11 is around the corner."
All those technologies failed miserably in the marketplace-- yet, Apple is still around and doing quite nicely in its niche market. Commodore bit the big one ages ago (in web weeks), yet the Amiga still has a dedicated hardcore cult following. Harley-Davidson bikes are unreliable compared to rice rockets, yet there is still a subculture centered around them. Amiga enthusiasts are mostly in our 30s and 40s, so the AmigaOS still has about 40 years to get its act together before we all croak.
Unfortunately they chose QNX as their kernel, which is not only proprietary, but also has few fanatical supporters. (unlike either *BSD or Linux, both of which have lots of fanatical supporters.)The Amiga has plenty of its own fanatical supporters, thank-you-very-much. Sure, Amiga on Linux would be the ultimate geek toy, but "BSD fanatic" is an oxymoron compared to "Amiga fanatic." We're the people who got our kicks going to Apple and PC stores just to watch the salespeople squirm when we started comparing them to Amigas.
What if MS were to apply the Apple Mac OS X model to Windows, using GNU Linux instead of BSD? Darwin is (nominally) open, yet Apple retains control through its Aqua/Carbon/Cocoa model. Let's call a theoretical GNU Linux layer "Dante" (similar to Darwin), the MS-Windows backward-compatibility emulation environment "Wintel" (like Classic) and the new part of the OS "Winux" (like Carbon/Cocoa/Aqua/X-Windows). Couldn't such an OS allow MS to retain its monopoly while effectively seizing control of Linux? Would this be a victory for Linux, or would Dante destroy everything penguinheads hold dear?