there are lots of people here saying things like "what about quark?" or some other app. "we won't be able to run quark! Mac's will become less usefull." the reasoning seems to go. but the whole point of Apple droping System 9 is to encourage (okay, strongly encourage) the software vendors to get their act together, and start producing software for the modern system - or at least that works under Classic. by getting the software makers to stop dragging their feet, Apple is making OS X much more useful (and usable), and doing a service to their users.
of course, there's the seperate problem of support for old peripherials like scanner drivers, but that's a different issue.
it's still a choice: don't upgrade. i know people still running Win95, because they've got no need for the new "features", and the older stuff is smaller and nicer to older hardware.
Apple computer today announced that Shake, a recently-aquired software product, will sell for half price on Apple's own computer systems. The savings is approximatly $5,000. In a seperate, unrelated announcement, Apple announced it would be releasing a new, multi-button mouse for use with Shake. The new mouse, initially available in a three-button model, will sell for $5,000.
yes, i've looked at both Intel's and AMD's 64-bit architectures, as well as being familiar with several other 64-bit architectures.
The IA-64 is an entirely new architecture, it has no support for IA-32 instructions.
ah, good catch; i was being unclear. yes, the instruction sets are different. but all the IA-64 chips announced to date retain 386 compatability. that includes everything from MMX through Virtual 8086 and Real mode. ick. and because the IA-64 arch doesn't handle IA-32, as you note, the handling of compatability becomes very hairy. switches between modes are non-trivial, and duplicates most of the work of booting a '386. IA-64 doesn't maintain the IA-32 crap: Itanium and friends do.
...the processor in the PlayStation 2 is an example [of a commercially successful VLIW machine]
hmm. MIPS cores, while arguably VLIW, are much closer to traditional RISC machines than Intel's EPIC, for sure. i'd also note that the MIPS chip in the PS2 has a much more limited function than what Intel's aiming at. VLIW's a murky term, and i wasn't including MIPS chips, although that's a reasonable thing to do, in which case, yes, i stand corrected.
AMD's design is more conservative, less innovative, and less exciting. the handling of 32-bit code complicates the instruction set over Intel's 64-bit-only model, but dramatically simplifies the resulting chips. AMD's design is also more predictable. it's more familiar, and dramatically less work to write for - particularly to get anything resembling reasonable performance.
all this being said, i hope to be proven wrong. i'd love to kiss IA-32 binaries goodbye, and find Intel (or someone) producing an IA-64-only chip with no support for the legacy crud. but i'm just not convinced that IA-64 is what's going to make this happen. Intel's taken too many chances at once, and pushed too much of the problem onto the software writer's shoulders.
the fact that there's so much wrong with the chips (and cores) comming from both AMD and Intel just re-affirms by stance that Apple - and everyone else - should avoid that whole mess. PowerPC, Alpha, SPARC, MIPS, and POWER all have 64-bit down better than either of the upcomming solutions.
...for the near term, loyal mac users will stay with "known good" hardware.
maybe. but that pretty much sacrifices any chance of growing their hardware/systems market. also, for X-386 to cut against MS's cash flow would require Apple to cut into the bundled licensing deals with people like Gateway and Dell - and we all know how that goes. MS exerts significant preasure to prevent even dual-booting scenarios (note the Be lawsuit), and to cut into MS's revenue stream, it'd have to be single boot, with no Windows. simply not gonna happen. also, as you suggest, the sustainability of Apple's hardware business in such a new model is only "near term". it's highly unlikely that it'd last against the marketing powers of folks like Gateway and Dell, if they start pushing OS X compatable machines. the vast majority of people will take the $200 price cut for the generic system. i've got a few x86 boxes i'd simply love to run OS X on. it's a beautiful (aesthetically and from an engineering view) system. but i strongly hope Apple stays well away from that market, for their own good. at least for several years yet.
...and are basing your viewpoint entirely from what you've read people posting on/.,
if i though for a second that were true, i'd turn off my computer forever.
...or you really don't have a clue about chip design.
well, i fully admit to not being an electrical engineer, and not being an expert on any chip design, but i certainly "have a clue".
i'll agree fully that IA-64 is provocative, unusual, and arguably innovative. but none of those make it good. i'm aware of the unusual design decisions Intel's made with IA-64, but you have to at least concede the complexity of the design is way above AMD's 64-bit design (or pretty much every other 64-bit design). the compatability with multiple previous existing instruction sets is killer, for one. and the added burden their model will put on compiler writers (to get even reasonable performance) is tremendous. and for speculative gains - no VLIW design's ever performed like the theorists hope. AMD's design is more conservative, to be sure, and less "exciting", but much easier to write for, much simpler. what is it you find "crappy" about it?
//The POWER4 instruction set is *identical* to a PowerPC instruction set.
are you sure? it this is true, i'd be quite impressed (and pleasantly suprised). i know the original PowerPC arch started as a fork of the POWER development, simplifying the instruction set significantly (and helping IBM retain much of its flexability for future development). it'd be pretty impressive if IBM has merged its POWER architecture with the PowerPC one. can you provide some reference for this?
note also that (i think) the altivec stuff isn't even technically part of the PowerPC instruction set, but is defined as an extention.
IA-32 is a pretty poor design, compared to most other chip designes currently on the market. IA-64 maintains all its crap, adds HP legacy support (basically both as seperate cores), and adds the new IA-64 VLIW/EPIC instructions on top of that. VLIW's never seen a commercially succesfull implementation, nor one that performs anywhere close to more traditional processor designs. nobody's ever thrown as much time or money at it as Intel is, so it'll be interesting to see if they can keep it competative, but that doesn't change the design issues. and EPIC is a particularly risky (although interesting, from a research point of view) implementation of VLIW. the whole thing's a mess, from top to bottom. and that's before even talking about what the design's going to force on compiler writers. ick.
anything would be better than that, but AMD's 64-bit chip is actually not to shabby on its own right. it still keeps the IA-32 legacy crap, which i think is bad (although understandable), but is much better in nearly all other respects. it doesn't improve the 32-bit portion, no, but all the 64-bit stuff is done much better.
mind you, i'd still rather have an Alpha, Sparc, MIPS, PowerPC, or POWER chip (in no particular order) inside my system.
Intel chips are still overpriced for what you're getting. anyway, the suggestion of supporting both is interesting, but has proven several times to be commercially impractical. the most obvious example being WinNT. the most popular platform becomes the focus, the others are neglected. the focus for the OS and systems designers, but - much more importantly - the focus for the app designers. WinNT for MIPS and PowerPC were canned primarially because there were virtually no apps for them. mind you, OS X is actually in a much better place for this than NT was/is. OS X has some nice features for multi-platform binaries, and (assuming Apple is still building Macs, rather than running on commodity hardware, which is all but impossable) the difference won't start out nearly as great as they were between PCs and Alpha+MIPS+PowerPC workstations. but even so, it doubles the amount of testing and integration work for the app writers, and doubles pretty much the entire body of work for the OS writers. two-platform binaries would be almost twice as large as their specialized cousins, and may or may not come with a startup performance hit (depending on how OS X implements this feature - i don't know).
and, most importantly, for what gain? how much does the use of a PowerPC processor really add to the total price of the system? i can't find prices right now, but i can't imagine it being much more than ~$100. is that really worth it to Apple, or their customers? not likely.
god, not the IA-64. lets keep that confined to the windows ghetto as much as possible. it's about the worst chip design i've seen, ever. AMD's 64-bit design is way better. but even there, i'm not clear why they'd go with that vs. the PowerPC chips (64-bit versions exist, they know PowerPC works for them, and have no conversion) or moving to IBM's POWER4 (64-bit, proven history, similar to PowerPC for a simpler transition). if Apple's going anywhere off PowerPC, i'm betting on IBM.
When Be was concentrating on building hardware......and software to match, they were in good shape.
oh? by what measure? how many people bought BeBoxes? i did, but i've only ever met two other owners, and i work with geeks. sure, Be managed to hang on long enough to reach IPO, when the dot-com bubble was still inflating, but you've got alot of work to do to demonstrate that they were doing well by any means.
given the choice of one or the other, bluetooth/wifi is definatly preferable to a micro-CDRW. ignoring questions of how useful a micro-CDRW is (compared to a normal CD[RW]), you can use wireless to get to such devices in other systems; you cannot go the other way around. the wireless access also has much broader uses; the CD-style devices are useless for anything interactive, for example.
oh, right. port the kernel. great idea. let's get minix up on this thing as quick as possible. er, wait, did you mean that other kernel? well, okay, if you want ultrix instead. honestly, there's a pair of kernels i'd like to see running on this thing; i'm pretty sure neither are the one you meant (Plan 9 and Inferno). there is more to the world than Linux.
anyway, porting the kernel (any kernel running on 386 chips) isn't likely to be the big effort - dirvers are. based on the (minimal) tech info on their site, it sounds like they're using a "special" display and touch screen, and they may or may not be keen on giving out info on those. plus whatever other funky stuff they've done to the AT spec.
as you pointed out, these guys left Apple to build a Windows XP device. that says amazingly bad things to me about their understanding of building real systems (as opposed to hardware boxes), which is an important and overlooked skill in general that becomes even more important when dealing with handheld and/or specialty systems.
i've used it, but perhaps not as extensivly as you'd like for me to report on. i've had no issues. i've used it with several clients (Solaris, Plan 9, Inferno, and W2K) without issue, but havn't ever tried ssh tunneling.
on the multiple versions front, let me clarify. there's a version that does only X11 (built from the common Unix source), and there's a Mac OS X version that does Aqua, thus including X11. i only mentioned them both because the Aqua one seems to be currently unavailable.
Unix from AT&T had a similar problem. they kept calling it "System N" and incrementing N. when they hit "System V" (the first to use a roman numeral, i think), they started pushing the tagline "consider it standard" to combat the fears of people noting that there was no official standard for Unix, unlike pretty much everything else the Bell System did. and, lo and behold, people sorta took AT&T's word for it! the problem was, it worked so well they couldn't increment N any more, because "System V" was the standard (and thus whatever came next wouldn't be). so the went to "System V Release 2" and got through about SVR4 before that got "stuck" as a "standard" with the various splits and unification efforts in the unix community. so they did "SVR4.1" and so on. then they started adding letters when that got stuck. many commercial unixes will still tell you they run a "SVR4.3x" (i think) kernel. it wasn't until SCO (the then-owners of the original Unix strain) did UnixWare 7 (i think) that they incremented to SVR5.
can't help with the first and third points, but i'm not sure exactly what you're talking about with the second. are you looking for a specific remote desktop display? VNC, an efficient cross-platform remote desktop program, is available for free. it's done by AT&T's UK research labs. there's a pair of clients available, and servers both for X11-only and native Aqua (although that one, at osxvnc.com, seems to have gone away; the vnc site has an email address for the author. there's another at mdimension.com that i've not sued, but should work fine.). check it out.
yup. makes you think you're in the wrong line of work, huh? "yeah, uh, you want desks. definatly gonna need desks. grey ones. oo, and round the edges. definatly. that'll be $5K, pleasethankyou."
you're possibly right, maybe given that as the alternative, a personal budget would be more cost-effective. but maybe not. still keep in mind that doing it per-person means that becomes a recurring cost, rather than a more-or-less one-time cost. and, as i said, the per-person budget adds moving and/or storage expenses. also, and perhaps more importantly, it's quite possible to simply not employ one of those stupid office specialist firms. when we laid out our first lab, we simply told an in-house HR person what we wanted, and he poked around in the basement and found stuff for us. much cheaper for the company, and we got good stuff (since he'd gotten input from us and was pretty sharp himself). the second time, we went down and roamed around the surplus area in the basement ourselves. not quite as cost effective for the company (the four of us combined, thankfully, got paid more than that one HR person), but still way better than the office specialist route, and much better than ther per-person shopping budget, and we got nearly exactly what we're looking for. now, that wouldn't work in smaller companies (since they're not likely to have a useful surplus of furniture lying around), but the first model's still reasonable.
well, for one thing, i was talking about the more general case. i have no interest in debating a given companies financials. and i wasn't actually aware of anyone mentioning MicroSoft. but more relavantly, don't underestimate how big the discounts you get for buying bulk are. a company with 10,000 is going to have a huge savings going with some uniform furniture, even if they give the employees a choice between a few styles.
that's a nice idea at first, but it's neither cost-effective for the company nor productive for the employee.
first, for the company: basic economic facts show that you do better buying in bulk. a company can order 1,000 office setups for much lower than it'd cost the workers to go out and buy it themselves. it's also a waste of time - most large companies have a HR person to do this, so i can focus on doing what i'm hired to do. and if the company simply divides what they normally spend on an office to the number of people in it, you're gonna get a real skimpy budget to work with. also note that, for the company, this would effectivly mean re-spending that money every time someone new shows up. i'm not likely to have the same tastes in furniture you do. and then, in addition to buying new furniture, they have to do something (store or dispose of) the old furniture. more cost. this'll all further impact the cost to the company and the budget you get to work with.
for the worker, a communal workplace can be much more productive than private (or pseudo-private) workplaces (and certainly far more productive than cubes). being able to sit right next to, with no barrier, the half-dozen people you're working on a project with is a huge win. being able to walk into a lab down the hall and find a half-dozen people working on another project is a big win for knowledge sharing, too.
also, i'm not sure what you're "I hate cubicles!" bit is refering to. nobody's advocated cubes. i agree, they're awful. but that's not what's being discussed. the problem with cubicles is that they don't offer the privacy and chance for "personality" that offices do, but they cut you off from co-workers and provide enough uniformity to make you feel uncomfortable. the solution described eliminates the sense of being cut off from co-workers, so that you can talk to people you're working with more easially, and so you develop something of a sense of community.
oh, and threre's this great new gadget i ran across (maybe i should post a slashdot article suggestion) for listening to music in shared spaces: headphones! or even just a volume knob.
the choice of what furniture and whatnot to stick at each workstation remains an open question.
I worked in a lab at Bell Labs where we formed a setup somewhat similar to this. we wanted it, and had to fight some to keep it when we moved, but we really like the model. basically, we had one large lab that we all spent the majority of our time in. before our move, everyone had an office, as well, but when that corridor was torn down for renovations and we had to move, we gave up our offices [1] to be allowed to tear down walls in the spot we were moving into to turn three mid-sized rooms into one large one. we set up desks with Plan 9 terminals (where there's nothing but a boot loader and swap space on local disk) around the perimiter of the room, a nice conference table in the middle, and a television (for demos and such) and couch at one end. [1] for a group of 6-8 people, we kept two offices. this was really useful for when you were doing work that required serious concentration, or if you had to make a private phone call. they were seldom used.
the setup worked great. there were typically 4-6 people in the lab. the theory of being able to pick any terminal did slowly evolve into people having a "normal" terminal, where people would leave manuals and such. evey once in a while, you'd walk in and there'd be someone passed out on the couch (if you were a loud sleeper, you used one of the offices, which also had a couch in it). we had one guy who telecumuted from halfway across the country, and was on-site for one week a month; after the second visit or so, he stopped getting a hotel and lived out of the office (thankfully, we had showers in the basement). whenever someone walked in who wasn't normally there who wanted to show/ask us something, or who we wanted to show/ask something, she'd just log in and go.
your reaction was not unheardof among people neighboring our odd lab. but you're describing very much a worst-case scenario, and is impacted by a number of factors, probably most strongly by how management (we pretty much managed ourselves, basically appointing one of us as an acting manager) treats it. in our lab, personal decorations were encouraged (everyone was expected to bring at least something in, and people hung a flag of choice around the perimiter of the ceiling - the lab quickly got the name "the Flag Room"). management needn't force people to move around, but should encourage people to sit by whoever they're working with at the moment. if you work with the same group of people on a long project, it makes sense you'll sit near each other for a good while. management should tell people not to do anything that prevents people from logging in at that terminal, though.
the results of this sort of environment are that you form a stronger community with your co-workers, and you get all the beifits that go along with it. your code (or other work) is of better quality, because it's dramatically easier to say "hey, what do you think of this?" than it is when you even have to go to the cube next door. the two private offices recognize the occasional need for real privacy (which cubes don't give at all, despite the illusion of). you form a social bond with the people you work with, too, which is just sort of a nice side-benifit. and you tend to be right next to the people you're actively working with, moving to another lab to work with a different set of people (although there were only two like this in the building i know of).
//Yes, DARPA had one really great hit [the internet] -- about 34 years ago.
let's assume for a moment that you're correct in your assertion that the internet was [d]arpa's only big win. well, hey, that's pretty big, no? and work on it certainly didn't stop 34 years ago. having the physical network running and off the ground was by no means the end of the arpanet, later the internet, project. bind, for example, was a direct result of [d]arpa funding. sure, it's not the best DNS resolver or server out there, but it does the job, and was crucial early on. [d]arpa funding for internet-related work continued for quite some time, and may still today (they've got an impressive list of projects and i didn't feel like reading through them all just to write this response).
further, and more importantly, your assertion is just false. while the internet work is certainly their highest-profile win, it's by no means the only project that's had a direct impact on our lives. all manner of research into communications systems, rocetry and propulsion, and materials sciences has led to the possability of orbital communications satalites (not to mention the more direct military applications). much advanced work in robotics, which has led to serious changes in manufacturing processes (including lowering costs for consumers and improving quality of goods), as well as (again) military and pseudo-military applications like bomb detection/disarmament. funding work in semiconductor and integrated circuit research led to the first SUN workstation at Stanford (that's what the S initially stood for) and dramatically influenced microprocessor design through today via the creation of the MIPS architecture and significant advancment of RISC architectures in general. recent research in computer vision and AI systems are resulting in unexpected gains in security applications in the private sector. current ongoing research in human vision, while already proving productive in military applications, looks quite promising for restoring normal or near-normal sight to people who suffer from various forms of serious debilitating vision impairment in the next few years.
i could go on, but i think you get the point. you wanna talk about beurocracy, fine. i agree there's likely problems that could be reasonably addressed. but DARPA remains one of the better ways the U.S. federal government spends taxpayer money. DARPA projects save lives and improve the quality of life for millions, in America and abroad. don't sell them short.
Slashdot Mirror
there are lots of people here saying things like "what about quark?" or some other app. "we won't be able to run quark! Mac's will become less usefull." the reasoning seems to go. but the whole point of Apple droping System 9 is to encourage (okay, strongly encourage) the software vendors to get their act together, and start producing software for the modern system - or at least that works under Classic. by getting the software makers to stop dragging their feet, Apple is making OS X much more useful (and usable), and doing a service to their users.
of course, there's the seperate problem of support for old peripherials like scanner drivers, but that's a different issue.
it's still a choice: don't upgrade. i know people still running Win95, because they've got no need for the new "features", and the older stuff is smaller and nicer to older hardware.
Apple computer today announced that Shake, a recently-aquired software product, will sell for half price on Apple's own computer systems. The savings is approximatly $5,000.
In a seperate, unrelated announcement, Apple announced it would be releasing a new, multi-button mouse for use with Shake. The new mouse, initially available in a three-button model, will sell for $5,000.
hey, how's it my fault the build the cases out of such crappy plastic. if they care so much about me chiping them, just build the cases better!
AMD's design is more conservative, less innovative, and less exciting. the handling of 32-bit code complicates the instruction set over Intel's 64-bit-only model, but dramatically simplifies the resulting chips. AMD's design is also more predictable. it's more familiar, and dramatically less work to write for - particularly to get anything resembling reasonable performance.
all this being said, i hope to be proven wrong. i'd love to kiss IA-32 binaries goodbye, and find Intel (or someone) producing an IA-64-only chip with no support for the legacy crud. but i'm just not convinced that IA-64 is what's going to make this happen. Intel's taken too many chances at once, and pushed too much of the problem onto the software writer's shoulders.
the fact that there's so much wrong with the chips (and cores) comming from both AMD and Intel just re-affirms by stance that Apple - and everyone else - should avoid that whole mess. PowerPC, Alpha, SPARC, MIPS, and POWER all have 64-bit down better than either of the upcomming solutions.
also, for X-386 to cut against MS's cash flow would require Apple to cut into the bundled licensing deals with people like Gateway and Dell - and we all know how that goes. MS exerts significant preasure to prevent even dual-booting scenarios (note the Be lawsuit), and to cut into MS's revenue stream, it'd have to be single boot, with no Windows. simply not gonna happen.
also, as you suggest, the sustainability of Apple's hardware business in such a new model is only "near term". it's highly unlikely that it'd last against the marketing powers of folks like Gateway and Dell, if they start pushing OS X compatable machines. the vast majority of people will take the $200 price cut for the generic system.
i've got a few x86 boxes i'd simply love to run OS X on. it's a beautiful (aesthetically and from an engineering view) system. but i strongly hope Apple stays well away from that market, for their own good. at least for several years yet.
i'll agree fully that IA-64 is provocative, unusual, and arguably innovative. but none of those make it good. i'm aware of the unusual design decisions Intel's made with IA-64, but you have to at least concede the complexity of the design is way above AMD's 64-bit design (or pretty much every other 64-bit design). the compatability with multiple previous existing instruction sets is killer, for one. and the added burden their model will put on compiler writers (to get even reasonable performance) is tremendous. and for speculative gains - no VLIW design's ever performed like the theorists hope. AMD's design is more conservative, to be sure, and less "exciting", but much easier to write for, much simpler. what is it you find "crappy" about it?
//The POWER4 instruction set is *identical* to a PowerPC instruction set.
are you sure? it this is true, i'd be quite impressed (and pleasantly suprised). i know the original PowerPC arch started as a fork of the POWER development, simplifying the instruction set significantly (and helping IBM retain much of its flexability for future development). it'd be pretty impressive if IBM has merged its POWER architecture with the PowerPC one. can you provide some reference for this?
note also that (i think) the altivec stuff isn't even technically part of the PowerPC instruction set, but is defined as an extention.
uh, what?
IA-32 is a pretty poor design, compared to most other chip designes currently on the market. IA-64 maintains all its crap, adds HP legacy support (basically both as seperate cores), and adds the new IA-64 VLIW/EPIC instructions on top of that. VLIW's never seen a commercially succesfull implementation, nor one that performs anywhere close to more traditional processor designs. nobody's ever thrown as much time or money at it as Intel is, so it'll be interesting to see if they can keep it competative, but that doesn't change the design issues. and EPIC is a particularly risky (although interesting, from a research point of view) implementation of VLIW. the whole thing's a mess, from top to bottom. and that's before even talking about what the design's going to force on compiler writers. ick.
anything would be better than that, but AMD's 64-bit chip is actually not to shabby on its own right. it still keeps the IA-32 legacy crap, which i think is bad (although understandable), but is much better in nearly all other respects. it doesn't improve the 32-bit portion, no, but all the 64-bit stuff is done much better.
mind you, i'd still rather have an Alpha, Sparc, MIPS, PowerPC, or POWER chip (in no particular order) inside my system.
Intel chips are still overpriced for what you're getting.
anyway, the suggestion of supporting both is interesting, but has proven several times to be commercially impractical. the most obvious example being WinNT. the most popular platform becomes the focus, the others are neglected. the focus for the OS and systems designers, but - much more importantly - the focus for the app designers. WinNT for MIPS and PowerPC were canned primarially because there were virtually no apps for them.
mind you, OS X is actually in a much better place for this than NT was/is. OS X has some nice features for multi-platform binaries, and (assuming Apple is still building Macs, rather than running on commodity hardware, which is all but impossable) the difference won't start out nearly as great as they were between PCs and Alpha+MIPS+PowerPC workstations.
but even so, it doubles the amount of testing and integration work for the app writers, and doubles pretty much the entire body of work for the OS writers. two-platform binaries would be almost twice as large as their specialized cousins, and may or may not come with a startup performance hit (depending on how OS X implements this feature - i don't know).
and, most importantly, for what gain? how much does the use of a PowerPC processor really add to the total price of the system? i can't find prices right now, but i can't imagine it being much more than ~$100. is that really worth it to Apple, or their customers? not likely.
god, not the IA-64. lets keep that confined to the windows ghetto as much as possible. it's about the worst chip design i've seen, ever. AMD's 64-bit design is way better. but even there, i'm not clear why they'd go with that vs. the PowerPC chips (64-bit versions exist, they know PowerPC works for them, and have no conversion) or moving to IBM's POWER4 (64-bit, proven history, similar to PowerPC for a simpler transition). if Apple's going anywhere off PowerPC, i'm betting on IBM.
given the choice of one or the other, bluetooth/wifi is definatly preferable to a micro-CDRW. ignoring questions of how useful a micro-CDRW is (compared to a normal CD[RW]), you can use wireless to get to such devices in other systems; you cannot go the other way around. the wireless access also has much broader uses; the CD-style devices are useless for anything interactive, for example.
// If the screen's that large, where are you going to put the controls?
use a touchscreen.
someone's done this already, forget who. Samsung?
//...port the kernel over to it?
oh, right. port the kernel. great idea. let's get minix up on this thing as quick as possible. er, wait, did you mean that other kernel? well, okay, if you want ultrix instead.
honestly, there's a pair of kernels i'd like to see running on this thing; i'm pretty sure neither are the one you meant (Plan 9 and Inferno). there is more to the world than Linux.
anyway, porting the kernel (any kernel running on 386 chips) isn't likely to be the big effort - dirvers are. based on the (minimal) tech info on their site, it sounds like they're using a "special" display and touch screen, and they may or may not be keen on giving out info on those. plus whatever other funky stuff they've done to the AT spec.
as you pointed out, these guys left Apple to build a Windows XP device. that says amazingly bad things to me about their understanding of building real systems (as opposed to hardware boxes), which is an important and overlooked skill in general that becomes even more important when dealing with handheld and/or specialty systems.
i've used it, but perhaps not as extensivly as you'd like for me to report on. i've had no issues. i've used it with several clients (Solaris, Plan 9, Inferno, and W2K) without issue, but havn't ever tried ssh tunneling.
on the multiple versions front, let me clarify. there's a version that does only X11 (built from the common Unix source), and there's a Mac OS X version that does Aqua, thus including X11. i only mentioned them both because the Aqua one seems to be currently unavailable.
Unix from AT&T had a similar problem. they kept calling it "System N" and incrementing N. when they hit "System V" (the first to use a roman numeral, i think), they started pushing the tagline "consider it standard" to combat the fears of people noting that there was no official standard for Unix, unlike pretty much everything else the Bell System did. and, lo and behold, people sorta took AT&T's word for it!
the problem was, it worked so well they couldn't increment N any more, because "System V" was the standard (and thus whatever came next wouldn't be).
so the went to "System V Release 2" and got through about SVR4 before that got "stuck" as a "standard" with the various splits and unification efforts in the unix community. so they did "SVR4.1" and so on. then they started adding letters when that got stuck.
many commercial unixes will still tell you they run a "SVR4.3x" (i think) kernel. it wasn't until SCO (the then-owners of the original Unix strain) did UnixWare 7 (i think) that they incremented to SVR5.
just a bit of Unix history.
can't help with the first and third points, but i'm not sure exactly what you're talking about with the second. are you looking for a specific remote desktop display? VNC, an efficient cross-platform remote desktop program, is available for free. it's done by AT&T's UK research labs. there's a pair of clients available, and servers both for X11-only and native Aqua (although that one, at osxvnc.com, seems to have gone away; the vnc site has an email address for the author. there's another at mdimension.com that i've not sued, but should work fine.). check it out.
//Have you seen what those... places charge...
yup. makes you think you're in the wrong line of work, huh? "yeah, uh, you want desks. definatly gonna need desks. grey ones. oo, and round the edges. definatly. that'll be $5K, pleasethankyou."
you're possibly right, maybe given that as the alternative, a personal budget would be more cost-effective.
but maybe not. still keep in mind that doing it per-person means that becomes a recurring cost, rather than a more-or-less one-time cost. and, as i said, the per-person budget adds moving and/or storage expenses.
also, and perhaps more importantly, it's quite possible to simply not employ one of those stupid office specialist firms. when we laid out our first lab, we simply told an in-house HR person what we wanted, and he poked around in the basement and found stuff for us. much cheaper for the company, and we got good stuff (since he'd gotten input from us and was pretty sharp himself). the second time, we went down and roamed around the surplus area in the basement ourselves. not quite as cost effective for the company (the four of us combined, thankfully, got paid more than that one HR person), but still way better than the office specialist route, and much better than ther per-person shopping budget, and we got nearly exactly what we're looking for.
now, that wouldn't work in smaller companies (since they're not likely to have a useful surplus of furniture lying around), but the first model's still reasonable.
well, for one thing, i was talking about the more general case. i have no interest in debating a given companies financials. and i wasn't actually aware of anyone mentioning MicroSoft. but more relavantly, don't underestimate how big the discounts you get for buying bulk are. a company with 10,000 is going to have a huge savings going with some uniform furniture, even if they give the employees a choice between a few styles.
//Do you know what a Shift key is?
:-)
yup. it's that thingie that lets me type double quotes, parens, and the like. why, whadda you use it for?
that's a shame; it really is quite nice. VGA card issues? that's the most common problem, by far.
that's a nice idea at first, but it's neither cost-effective for the company nor productive for the employee.
first, for the company: basic economic facts show that you do better buying in bulk. a company can order 1,000 office setups for much lower than it'd cost the workers to go out and buy it themselves. it's also a waste of time - most large companies have a HR person to do this, so i can focus on doing what i'm hired to do. and if the company simply divides what they normally spend on an office to the number of people in it, you're gonna get a real skimpy budget to work with.
also note that, for the company, this would effectivly mean re-spending that money every time someone new shows up. i'm not likely to have the same tastes in furniture you do. and then, in addition to buying new furniture, they have to do something (store or dispose of) the old furniture. more cost. this'll all further impact the cost to the company and the budget you get to work with.
for the worker, a communal workplace can be much more productive than private (or pseudo-private) workplaces (and certainly far more productive than cubes). being able to sit right next to, with no barrier, the half-dozen people you're working on a project with is a huge win. being able to walk into a lab down the hall and find a half-dozen people working on another project is a big win for knowledge sharing, too.
also, i'm not sure what you're "I hate cubicles!" bit is refering to. nobody's advocated cubes. i agree, they're awful. but that's not what's being discussed. the problem with cubicles is that they don't offer the privacy and chance for "personality" that offices do, but they cut you off from co-workers and provide enough uniformity to make you feel uncomfortable. the solution described eliminates the sense of being cut off from co-workers, so that you can talk to people you're working with more easially, and so you develop something of a sense of community.
oh, and threre's this great new gadget i ran across (maybe i should post a slashdot article suggestion) for listening to music in shared spaces: headphones! or even just a volume knob.
the choice of what furniture and whatnot to stick at each workstation remains an open question.
I worked in a lab at Bell Labs where we formed a setup somewhat similar to this. we wanted it, and had to fight some to keep it when we moved, but we really like the model. basically, we had one large lab that we all spent the majority of our time in. before our move, everyone had an office, as well, but when that corridor was torn down for renovations and we had to move, we gave up our offices [1] to be allowed to tear down walls in the spot we were moving into to turn three mid-sized rooms into one large one. we set up desks with Plan 9 terminals (where there's nothing but a boot loader and swap space on local disk) around the perimiter of the room, a nice conference table in the middle, and a television (for demos and such) and couch at one end.
[1] for a group of 6-8 people, we kept two offices. this was really useful for when you were doing work that required serious concentration, or if you had to make a private phone call. they were seldom used.
the setup worked great. there were typically 4-6 people in the lab. the theory of being able to pick any terminal did slowly evolve into people having a "normal" terminal, where people would leave manuals and such. evey once in a while, you'd walk in and there'd be someone passed out on the couch (if you were a loud sleeper, you used one of the offices, which also had a couch in it). we had one guy who telecumuted from halfway across the country, and was on-site for one week a month; after the second visit or so, he stopped getting a hotel and lived out of the office (thankfully, we had showers in the basement). whenever someone walked in who wasn't normally there who wanted to show/ask us something, or who we wanted to show/ask something, she'd just log in and go.
your reaction was not unheardof among people neighboring our odd lab. but you're describing very much a worst-case scenario, and is impacted by a number of factors, probably most strongly by how management (we pretty much managed ourselves, basically appointing one of us as an acting manager) treats it. in our lab, personal decorations were encouraged (everyone was expected to bring at least something in, and people hung a flag of choice around the perimiter of the ceiling - the lab quickly got the name "the Flag Room"). management needn't force people to move around, but should encourage people to sit by whoever they're working with at the moment. if you work with the same group of people on a long project, it makes sense you'll sit near each other for a good while. management should tell people not to do anything that prevents people from logging in at that terminal, though.
the results of this sort of environment are that you form a stronger community with your co-workers, and you get all the beifits that go along with it. your code (or other work) is of better quality, because it's dramatically easier to say "hey, what do you think of this?" than it is when you even have to go to the cube next door. the two private offices recognize the occasional need for real privacy (which cubes don't give at all, despite the illusion of). you form a social bond with the people you work with, too, which is just sort of a nice side-benifit. and you tend to be right next to the people you're actively working with, moving to another lab to work with a different set of people (although there were only two like this in the building i know of).
//Yes, DARPA had one really great hit [the internet] -- about 34 years ago.
let's assume for a moment that you're correct in your assertion that the internet was [d]arpa's only big win. well, hey, that's pretty big, no? and work on it certainly didn't stop 34 years ago. having the physical network running and off the ground was by no means the end of the arpanet, later the internet, project. bind, for example, was a direct result of [d]arpa funding. sure, it's not the best DNS resolver or server out there, but it does the job, and was crucial early on. [d]arpa funding for internet-related work continued for quite some time, and may still today (they've got an impressive list of projects and i didn't feel like reading through them all just to write this response).
further, and more importantly, your assertion is just false. while the internet work is certainly their highest-profile win, it's by no means the only project that's had a direct impact on our lives. all manner of research into communications systems, rocetry and propulsion, and materials sciences has led to the possability of orbital communications satalites (not to mention the more direct military applications). much advanced work in robotics, which has led to serious changes in manufacturing processes (including lowering costs for consumers and improving quality of goods), as well as (again) military and pseudo-military applications like bomb detection/disarmament. funding work in semiconductor and integrated circuit research led to the first SUN workstation at Stanford (that's what the S initially stood for) and dramatically influenced microprocessor design through today via the creation of the MIPS architecture and significant advancment of RISC architectures in general. recent research in computer vision and AI systems are resulting in unexpected gains in security applications in the private sector. current ongoing research in human vision, while already proving productive in military applications, looks quite promising for restoring normal or near-normal sight to people who suffer from various forms of serious debilitating vision impairment in the next few years.
i could go on, but i think you get the point. you wanna talk about beurocracy, fine. i agree there's likely problems that could be reasonably addressed. but DARPA remains one of the better ways the U.S. federal government spends taxpayer money. DARPA projects save lives and improve the quality of life for millions, in America and abroad. don't sell them short.