But if you paid $2000 for a laptop, or $300 for a PDA, wouldn't you take care of the damn thing?
Sure, but things happen. I had a Visor (palmOS) that fell to the concrete while I was snapping the cover back on. I had been running and my hands were too sweaty. That didn't happen to break it, but it could have. Same one a year later I did manage to break. I had it out to turn off an alarm. I didn't put it back on the coffee table, but left it on the floor instead. The dog came over with a toy and we played a while, and I managed to put my knee on it. Oops, $80 to repair the screen.
Oh, that doesn't count the time I fell in a hot tub, but I didn't happen to have the PDA on me, just the cell phone...
If you have the thing with you every day for a large part of the day, it is going to be in a tad bit of danger now and again. It is nice if they are either rugged enough to handle it, or cheap enough to fix/replace when they do get killed.
Plus, I would love a supper rugged laptop, I could take it into the hot tub and read (photography) news while working out the day's stress (I wouldn't want to read the other kind of news...it increases stress).
There are also significant differences in the boot procedure (one of the things that I prefer about SysV). BSD has one file (script) per runlevel. SysV has one script per service, organized in 1 directory per runlevel. Want to stop a service in sysv? ' stop'.
FYI, NetBSD has the script per-service (incl ' stop', and ' status') scheme. FreeBSD is experimenting with it as well (but have not decided for sure if they should adopt it). There is a Usenix paper about it, try the 2001 procedings.
Neither has the concept of runlevels though, other then single-user and multi-user that is.
The features just don't compare. The iMac has a ton of stuff that you're not getting in a $500 PC, everything from hardware features, to bundled software, to future support. Integration, ease of use. Mac OS X is no Windows Me, either.
What kind of hardware features? I don't have an iMac, but the only thing I can think of that they have an entry level PCs might not is CD-RW drives. Oh, and the monitor is much nicer on the iMac, but people still look at them both as into level computers and see the price difference.
If the user chooses, Macs update all of their software components automatically once a day/week/month. You leave an iMac alone and your video drivers, security patches, OS updates, updates to bundled apps, etc. all just show up on their own, popping up to ask for your approval to install, with descriptions of what they do, all vetted by Apple. Then major OS updates are one-click installers that just work, without driver issues and long serial numbers and activation, and every other one is free. You don't get stuff like that in a $500 PC. It costs money to develop that kind of thing. There are dozens of features like that in an iMac or any Mac.
I'm not a big windows user, but is this really different from the Windows Updater? My PCs pretty much just run Unix, so I have little basis to compare to my Mac.
Apple uses only its own propietary hardware. The hardware support is minimal. Meanwhile, Windows has to support a plethora of parallel, serial, USB, IEEE1394 devices, PCI, AGP, ISA devices.
Most of the Windows drivers are provided by the device maker. I think Apple provides more 1394 drivers then MS does -- Adaptec (and others) do the 1394 for MS for free, while Apple does their own for example.
I don't think MS has done any AGP drivers, the video card makers do it. I don't know how many video drivers Apple actually does, the video card makers may also do theirs, and if not there are only 3 or so a year tops.
I am pretty sure Apple had to write a fair number of USB drivers on their own...
The only "battery thing" I saw was that it's rechargable. Big whoop. Palms have had that in units for quite some time. Come to think of it, even the Visor Prism has it.
The "problem" with the Prism and Palm V and others is the battery takes a while to charge. Long enough that you are recommended to leave it one the charger overnight.
With the Edge (and I assume the newer Neo, and whatever the other one is called) it tops off in a few minutes, so a normal hotsync should keep it charged. I think about half an hour will do a full charge.
The downside is they don't suck the power off the USB, they have a power dongle that plugs into the USB. So it is a pain to deal with if you want to sync to a laptop or something and also charge it. It also doesn't come with a travel charger or anything, so the only way to top it off is to drag along the whole sync cradle and power wart and all. Of corse I think the point is you don't need to travel charger unless you take trips longer then a month...
I don't see much reason to get one of these vs. the Edge, unless I guess you are always going to have a module in it (the Edge isn't very thin if you have to slap on the springboard slot), so I guess if you are going to make full time use of the GPS module, or the 802.11 it makes more sense to get one of the newer ones.
The only reason I preferred the Handspring to the Palm was that it took AAA batteries. I have yet to encounter a recharger while camping (and no, I can't plug it into a "current" bush.)
How long do you camp for? I don't know about the newer Handsprings, but the Edge's rechargable lasts me about a month if I'm careful not to charge it up! (careful being not plugging in the charger)
Darwin and OS X do not use BSD device drivers. They use a completely different driver architecture. Common wisdom on darwin-development is that it is easier to rewrite a BSD or Linux driver for Darwin than it is to port it.
From what I understood there is a BSD compatability layer to let you use unmodifyed BSD device drivers (so long as they use the bus_space code, and have no x86 assembly), and also a way to use either a C++ or ObjectaveC subset. Clearly they are promoting use of the new OO way, otherwise they wouldn't have bothered to provide it.
I havn't personally done either since Apple kindly produced drivers for everything my notebook has except the DVD video.
The problem is that Apple never lowered prices even back when they had much larger than 5% market share. They seem to have long ago decided on a pricing structure that has settled them into a 5% market of loyal users, and they must figure this maximizes profits for them. Increasing market share by lowering prices doesn't seem to be (and never has been) an attractive strategy for Apple, and, as you mention, the more they let their market share slide, the harder it is to do.
It is hard to disagree. However i think last time around they were more costly because they were doing things like using SCSI across the line. Maybe next time around (if there is one!) they will try harder. Or maybe not.
They may have been on the right track with the i-Mac, but they didn't keep up the push by upgrading rapidly and continuing to reduce prices, and that one too has languished. It's really a shame--at one point i-Macs were flying off the shelves nearly as fast as Wintel hardware. I had a lot of hope for Apple at that moment.
You are right about the iMac, but I think it is less that they have failed to follow through, as Intel and the mobo makers have rushed to fill in the new niche Apple "discovered", and Apple has a hard time fighting that. I mean today's iMac really is nicer then the original by a fair margin, but the prices haven't fallen (they have gone up a little even), so now they are way behind the $500 Wintel box, the box that didn't really exist when iMac first came out (not $500 with a monitor at least).
As someone else mentioned, G4/G5 PPC machines may be a bit better than Intel PCs, but will most people perceive them as being worth nearly _twice_ as much? Whenever I've been in the market for a new PC, I've always checked out the current crop of Mac hardware. Each time, I have liked what I've seen, but simply could not justify paying almost twice as much for similar or at most slightly better performance.
Yep. The only machines I see that are price competitive are their laptops, which are selling very well at the moment. It may not be long until PC laptops pull ahead again though.
It does NOT cost as much to develop OSX as it took to develop W2K! Apple took the Code straight from the BSD and NextStep projects.
It cost a whole lot to buy NeXT in the first place... Plus they don't actually use a FreeBSD kernel, it is still MACH with a BSD compatability layer so they can use BSD device drivers (which they do -- where they are availble). I don't think they got anything from AfterStep either.
Apple saved a crapload of money thanks to the BSD guys, and quite frankly I'm still waiting to see Apple give anything substantial BACK to the BSD community.
Well, they did give back the whole lower layer of their OS (see Darwin). Also gcc changes, including the ability to use pre-parsed header files. The changes havn't been adopted back yet because they aren't "clean" enough. I do beleve Apple is going to try again, at least with the gcc changes.
You obviously arent aware that apple is pulling a 30% profit margin on every machine shipped.
Apple's products seem to cost a fair bit more then 30% more then other people's. Their entry level iMac is around $900, and entry level PC is around $500. The only real exception is the notebooks where they seem alot nicer for the money.
BTW, does the 30% "profit margin" mean 30% over the cost of goods, or 30% over cost of goods plus NREs?
"Doesn't allow OSS work" only means something if you're doing it during their time-
Bull. It depends on what you sign. I turned down several jobs this year because they wanted all intellectual property, at least one only wanted it in their field, but I didn't want to take the chance.
Sure that might not stand up in court, but it might.
Wow, with 1.6 Ghz ready in a few months and a possible version at 2.0 Ghz Apple might be able to drop the PPC Mhz is not a Intel Mhz campaign that they were doing a while ago.
Why? You think Intel is going to stand still as Motorola catches up?
Besides it is still likely to be true. You can pretty much never compare Mhz to Mhz between two different CPU designs and come out with the right idea. You really do need to benchmark what you will do with it. Of corse it could be that the P-IIII or K7 is faster, clock for clock, or only given it's higher clock rate depending on what you are running...
Apple's machines require much less cooling hardware, plus the PowerPC chips have fewer transistors and should be easier to produce in quantity. Most likely this is where Apple is making most of its money.
There are a lot of per product costs (aka non recurring expenses, or NREs). It costs roughly as much to develop a new version of MacOS as Windows. It costs roughly as much to design a new PowerPC as it does a new P-IIII or K7. Apple has about 5% of the market.
If you pretend it costs $100,000 to design a new OS and CPU, and that there are 100 people that buy computers, you can see that the 95 people who buy a Wintel box will have to pay about $1000 each for their share of the NRE. The 5 people that buy Apples have to pay about $20,000 each.
In the real world it isn't quite that bad since there are more uses for the PowerPC then just Apple's products. There are also more NREs that are similar in scale for PC makers. For example the video card in a Mac is pretty much just a PC video card. Apple ships about as many PCs as a big PC maker, so their cost to design a case and motherboard is about the same.
Still if Apple had 50% of the market rather then 5% they could manage to sell the machines for much closer to Wintel prices (maybe even under it).
I'm sure there are some other reasons, but I have a feeling that this is the biggest one...
Hmmm, maybe these new b0xen will actually be able to run OS X and all its interface dandies without feeling like you're on a an old 386.
Actually you can do that right now. I have a laptop, and OSX was pretty slugish on it. I bought another 512M of RAM (bringing it to 640M) and it runs much much much faster.
Granted that is pretty pigish, but at least you can fix it now. Apparently OS X 10.1 will be a lot faster as well (according to Apple at least), but we will know that later this month one way or another.
dont you just hate that? i really wish they would apply the karma over a rolling horizon.
Now there's an idea. Of course someone that posts a lot of things worth "2" will get enough karma to post at two by default, but will lose it because they don't post much worth "3"...
(of course the reality is good posts posted sooner get high ratings, good posts posted later don't go so high. I've had posts every bit as good as my 5's never go above 2 or 3 because they were posted late. Almost all of my fives were when the article it was attached too was in the top 3, normally the top most)
Seems reasonable to fire the people who would work on a given project for free anyways
If you have the lead developer for Project Foo, it helps you sell support contracts to places that are using Project Foo. If you don't have any developers for Project Foo it is way harder to sell support.
Seems reasonable to fire the people who would work on a given project for free anyways.
Yes, and no. You lose the ability to have influence over the direction of development. For example to decide which graphics cards get worked on first/next.
If the developer gets a new job that isn't to keep working on the code, they will also (probably) work on it less. So if VA needed the code to keep evolving then they are at risk there too.
There is also the chance that the new employer either doesn't allow OSS work, or has such an interesting project that the developer gives up on the old project...
Is there a 'contingency plan' if this horrible event were to occur?
Beats me, but the slash code is open source, and even though slashdot.org belongs to VA, one would assume it could go back to being chips & dips, and most readers would follow...
I'll lose my "low" user number, and high karma, but that doesn't really matter too much. I'm still in the process of losing my high karma here anyway:-) (I'm losing it mostly by having posts mod'ed up to 4 or 5, then as "overrated" back down to 2 or 3...)
Yes, key management is hard. But you have to do it.
You have to do it to be safe, you don't have to do it to sell -- otherwise 802.11 would have floped like IRDA.
I'm not convinced that the right way to do it is necessarily physical network element to physical network element. That has the advantage of being easy to understand, but the disadvantage that some elements are hard to reach.
One could also set up something where you authenticate to a device that can vouch for you to the rest of the network. That first step could be physical to physical, and the rest can be network authenticated (to the device you touched, or to something that it talks too). For example touching that first thing can make a KerbV principal, and then everything else uses plain old KerbV...
Of corse since that won't be hugely different from KerbV we would have to wonder why that would work when almost nothing uses KerbV right now despite free availability...
Rainbow, the leading maker of dongles, has acquired Mykotronx, which builds NSA-approved crypto. One result of this has been the iKey 2000 [rainbow.com] USB-based public key device. This is a key-sized device that plugs into a USB port and does public key encryption. It's not just a key - it's the crypto device itself. So the computer it's plugged into never sees the key. There's Windows support; someone might want to do Linux support
Cool, I did a similar thing with the Java iButton (which uses I2C rather then USB). It was really really slow since the Java on the iButton was quite slow.
Does Rainbow provide enough docs to do the host side for Linux (or *BSD)?
Neither 802.11 nor Bluetooth has a truly usable authentication scheme.
Pretty much, but it is less an issue for bluetooth since it has shorter range.
The way this ought to work is that wireless devices should have to be "introduced" before they talk. A separate "introduction interface", using short-range IR or physical contact, is needed. The idea is that if you want the printer to talk to a laptop, you point them at each other, they exchange keys, you punch a button on each saying you approve, and thereafter they can talk. Less-portable devices should be introduced through some intermediary, like a palmtop. Underneath is a crypto system, but the users never see it.
Pretty cool idea. There are some practical issues though. The best place for the 802.11 access point at my friends house is in the attic, only there does it get good covrage of the yard. The best place for it in my house is in the messy computer room, where I want very few guests to see. The best place at the last place I worked was in a locked room where the network came on to the floor (they are also centrally located there). Many of those places make physical contact hard or impossable.
It is also nice if you can roam from access point to access point, if you have to introduce yourself to each one it can be really confusing why one laptop has close to 100% signal, and another has barely 35% when they are less then three feet apart...
It is also nice to be able to replace a failed network part without everyone on the floor having to come shake hands with it (if they can even get to the thing).
On the other hand that would prevent people from connecting to the network just because they can get to the parking lot.
Of course, then you have the problem of your neighbours using your cable modem without your permission!
Yep. My network doesn't seem to leave my yard, but there there may be more sensitive 802.11 cards. I have dealt with it by having my dhcp server send me mail if it sees a new MAC address. If I ever get that mail, I can decide what to do about it.
I'm making the assumption that an "attacker" (or freeloader) would try the simple thing of just hooking up and trying DHCP.
Are you guys totaly out of touch with reality? There are even less people with a wireless network than with ethernet.
I may be out of touch with reality. Of all my friends with a ethernet network at home, they all also have a wireless network. Many of them have a wired network in only one or two rooms while the wireless covers the bulk of their houses (and part of the yards).
I know of a much smaller number of people with no wired network who are thinking of doing a wireless network.
My theory is that a wired network in one room is cheap and easy (the hardest part is getting an ISP to support it, or setting up NATing yourself). A wired network covering more then one room is normally kind of hard and/or costly (it is pretty easy and cheap if the house is still being built though!). A wireless network for a whole house is modestly expensive, but just as easy to install (plug a wireless card into all machines, plug in a access point).
So anyone with a single room network and get a whole house network quite easily by setting up 802.11. If you have a laptop it's a really easy choice:-) If you already have a multiroom network, I'm going to assume you are a true geek, and be quite intrested in having a wireless network as well. So both those cases make sense.
The no wired network, wireless only makes less sense. In part because it is so easy to do a one room network, and it is faster and somewhat safer. The one case where it makes sense is someone who can get a "cable modem" (or regular modem) into a room they don't want the computer in, and it is hard to get that access to the computer room. Then a wireless network might make more sense. I'm not sure people that don't want computers in every room really deserve wireless networks though:-)
Not really a downside if you use authentication and encryption at your access point. I think you're just picking at straws here!
Well I am picking at straws, it is my vocation, and my great skill. Oh, and I like it.
However if you are talking the 802.11 A&E, it doesn't seem to be so secure. Have you read slashdot recently?:-) Generically, yes, A&E solves all problems like that, but which A&E method? KerbV? SSH? SSL?
I think most 802.11 networks will be set up with no A&E, or with 802.11's WEP, which hasn't been all that successful. Those are the easy choices. Doing something else will be hard, nonstandard, and stand some chance of working.
(of corse easy and standard would be making the wireless network outside the normal firewall, but many places find that too inconvenient)
Easy solution to the security problem: VPN tunnel through the firewall. Both secure and useful.
I don't have an ethernet printer at home, but I do use SSH tunnels there (to get my mail, and run VNC, and/or X).
The last place I worked had an open 802.11 network, but it was treated as "outside the firewall" by everything at work. They also have another open 802.11 network in another building that is inside the firewall, which is a bad idea.
No, because ethernet ports are generally built into the mainboard, so would require a costly replacement of the hardware and firmware to be able to support Bluetooth, 802.11 or whatever.
I think the point is if you have a postscript printer on your IP network, and you hook up an 802.11 access point t your IP network, then your printer magically becomes an 802.11 printer as well (unless you fire wall off the 802.11 part, which is frequently a good idea, but not as frequently done).
The upside? One 802.11 access point gets all your printers. The downside? The longer range of 802.11 lets people outside your office waste your paper. Plus IP stacks on many printers are insanely fragile, so someone could break the printer pretty simply.
Slashdot Mirror
Sure, but things happen. I had a Visor (palmOS) that fell to the concrete while I was snapping the cover back on. I had been running and my hands were too sweaty. That didn't happen to break it, but it could have. Same one a year later I did manage to break. I had it out to turn off an alarm. I didn't put it back on the coffee table, but left it on the floor instead. The dog came over with a toy and we played a while, and I managed to put my knee on it. Oops, $80 to repair the screen.
Oh, that doesn't count the time I fell in a hot tub, but I didn't happen to have the PDA on me, just the cell phone...
If you have the thing with you every day for a large part of the day, it is going to be in a tad bit of danger now and again. It is nice if they are either rugged enough to handle it, or cheap enough to fix/replace when they do get killed.
Plus, I would love a supper rugged laptop, I could take it into the hot tub and read (photography) news while working out the day's stress (I wouldn't want to read the other kind of news...it increases stress).
FYI, NetBSD has the script per-service (incl ' stop', and ' status') scheme. FreeBSD is experimenting with it as well (but have not decided for sure if they should adopt it). There is a Usenix paper about it, try the 2001 procedings.
Neither has the concept of runlevels though, other then single-user and multi-user that is.
What kind of hardware features? I don't have an iMac, but the only thing I can think of that they have an entry level PCs might not is CD-RW drives. Oh, and the monitor is much nicer on the iMac, but people still look at them both as into level computers and see the price difference.
I'm not a big windows user, but is this really different from the Windows Updater? My PCs pretty much just run Unix, so I have little basis to compare to my Mac.
Most of the Windows drivers are provided by the device maker. I think Apple provides more 1394 drivers then MS does -- Adaptec (and others) do the 1394 for MS for free, while Apple does their own for example.
I don't think MS has done any AGP drivers, the video card makers do it. I don't know how many video drivers Apple actually does, the video card makers may also do theirs, and if not there are only 3 or so a year tops.
I am pretty sure Apple had to write a fair number of USB drivers on their own...
The "problem" with the Prism and Palm V and others is the battery takes a while to charge. Long enough that you are recommended to leave it one the charger overnight.
With the Edge (and I assume the newer Neo, and whatever the other one is called) it tops off in a few minutes, so a normal hotsync should keep it charged. I think about half an hour will do a full charge.
The downside is they don't suck the power off the USB, they have a power dongle that plugs into the USB. So it is a pain to deal with if you want to sync to a laptop or something and also charge it. It also doesn't come with a travel charger or anything, so the only way to top it off is to drag along the whole sync cradle and power wart and all. Of corse I think the point is you don't need to travel charger unless you take trips longer then a month...
I don't see much reason to get one of these vs. the Edge, unless I guess you are always going to have a module in it (the Edge isn't very thin if you have to slap on the springboard slot), so I guess if you are going to make full time use of the GPS module, or the 802.11 it makes more sense to get one of the newer ones.
How long do you camp for? I don't know about the newer Handsprings, but the Edge's rechargable lasts me about a month if I'm careful not to charge it up! (careful being not plugging in the charger)
From what I understood there is a BSD compatability layer to let you use unmodifyed BSD device drivers (so long as they use the bus_space code, and have no x86 assembly), and also a way to use either a C++ or ObjectaveC subset. Clearly they are promoting use of the new OO way, otherwise they wouldn't have bothered to provide it.
I havn't personally done either since Apple kindly produced drivers for everything my notebook has except the DVD video.
It is hard to disagree. However i think last time around they were more costly because they were doing things like using SCSI across the line. Maybe next time around (if there is one!) they will try harder. Or maybe not.
You are right about the iMac, but I think it is less that they have failed to follow through, as Intel and the mobo makers have rushed to fill in the new niche Apple "discovered", and Apple has a hard time fighting that. I mean today's iMac really is nicer then the original by a fair margin, but the prices haven't fallen (they have gone up a little even), so now they are way behind the $500 Wintel box, the box that didn't really exist when iMac first came out (not $500 with a monitor at least).
Yep. The only machines I see that are price competitive are their laptops, which are selling very well at the moment. It may not be long until PC laptops pull ahead again though.
It cost a whole lot to buy NeXT in the first place... Plus they don't actually use a FreeBSD kernel, it is still MACH with a BSD compatability layer so they can use BSD device drivers (which they do -- where they are availble). I don't think they got anything from AfterStep either.
Well, they did give back the whole lower layer of their OS (see Darwin). Also gcc changes, including the ability to use pre-parsed header files. The changes havn't been adopted back yet because they aren't "clean" enough. I do beleve Apple is going to try again, at least with the gcc changes.
Apple's products seem to cost a fair bit more then 30% more then other people's. Their entry level iMac is around $900, and entry level PC is around $500. The only real exception is the notebooks where they seem alot nicer for the money.
BTW, does the 30% "profit margin" mean 30% over the cost of goods, or 30% over cost of goods plus NREs?
Bull. It depends on what you sign. I turned down several jobs this year because they wanted all intellectual property, at least one only wanted it in their field, but I didn't want to take the chance.
Sure that might not stand up in court, but it might.
Why? You think Intel is going to stand still as Motorola catches up?
Besides it is still likely to be true. You can pretty much never compare Mhz to Mhz between two different CPU designs and come out with the right idea. You really do need to benchmark what you will do with it. Of corse it could be that the P-IIII or K7 is faster, clock for clock, or only given it's higher clock rate depending on what you are running...
There are a lot of per product costs (aka non recurring expenses, or NREs). It costs roughly as much to develop a new version of MacOS as Windows. It costs roughly as much to design a new PowerPC as it does a new P-IIII or K7. Apple has about 5% of the market.
If you pretend it costs $100,000 to design a new OS and CPU, and that there are 100 people that buy computers, you can see that the 95 people who buy a Wintel box will have to pay about $1000 each for their share of the NRE. The 5 people that buy Apples have to pay about $20,000 each.
In the real world it isn't quite that bad since there are more uses for the PowerPC then just Apple's products. There are also more NREs that are similar in scale for PC makers. For example the video card in a Mac is pretty much just a PC video card. Apple ships about as many PCs as a big PC maker, so their cost to design a case and motherboard is about the same.
Still if Apple had 50% of the market rather then 5% they could manage to sell the machines for much closer to Wintel prices (maybe even under it).
I'm sure there are some other reasons, but I have a feeling that this is the biggest one...
Actually you can do that right now. I have a laptop, and OSX was pretty slugish on it. I bought another 512M of RAM (bringing it to 640M) and it runs much much much faster.
Granted that is pretty pigish, but at least you can fix it now. Apparently OS X 10.1 will be a lot faster as well (according to Apple at least), but we will know that later this month one way or another.
Now there's an idea. Of course someone that posts a lot of things worth "2" will get enough karma to post at two by default, but will lose it because they don't post much worth "3"...
(of course the reality is good posts posted sooner get high ratings, good posts posted later don't go so high. I've had posts every bit as good as my 5's never go above 2 or 3 because they were posted late. Almost all of my fives were when the article it was attached too was in the top 3, normally the top most)
If you have the lead developer for Project Foo, it helps you sell support contracts to places that are using Project Foo. If you don't have any developers for Project Foo it is way harder to sell support.
Yes, and no. You lose the ability to have influence over the direction of development. For example to decide which graphics cards get worked on first/next.
If the developer gets a new job that isn't to keep working on the code, they will also (probably) work on it less. So if VA needed the code to keep evolving then they are at risk there too.
There is also the chance that the new employer either doesn't allow OSS work, or has such an interesting project that the developer gives up on the old project...
Beats me, but the slash code is open source, and even though slashdot.org belongs to VA, one would assume it could go back to being chips & dips, and most readers would follow...
I'll lose my "low" user number, and high karma, but that doesn't really matter too much. I'm still in the process of losing my high karma here anyway :-) (I'm losing it mostly by having posts mod'ed up to 4 or 5, then as "overrated" back down to 2 or 3...)
You have to do it to be safe, you don't have to do it to sell -- otherwise 802.11 would have floped like IRDA.
I'm not convinced that the right way to do it is necessarily physical network element to physical network element. That has the advantage of being easy to understand, but the disadvantage that some elements are hard to reach.
One could also set up something where you authenticate to a device that can vouch for you to the rest of the network. That first step could be physical to physical, and the rest can be network authenticated (to the device you touched, or to something that it talks too). For example touching that first thing can make a KerbV principal, and then everything else uses plain old KerbV...
Of corse since that won't be hugely different from KerbV we would have to wonder why that would work when almost nothing uses KerbV right now despite free availability...
Cool, I did a similar thing with the Java iButton (which uses I2C rather then USB). It was really really slow since the Java on the iButton was quite slow.
Does Rainbow provide enough docs to do the host side for Linux (or *BSD)?
Pretty much, but it is less an issue for bluetooth since it has shorter range.
Pretty cool idea. There are some practical issues though. The best place for the 802.11 access point at my friends house is in the attic, only there does it get good covrage of the yard. The best place for it in my house is in the messy computer room, where I want very few guests to see. The best place at the last place I worked was in a locked room where the network came on to the floor (they are also centrally located there). Many of those places make physical contact hard or impossable.
It is also nice if you can roam from access point to access point, if you have to introduce yourself to each one it can be really confusing why one laptop has close to 100% signal, and another has barely 35% when they are less then three feet apart...
It is also nice to be able to replace a failed network part without everyone on the floor having to come shake hands with it (if they can even get to the thing).
On the other hand that would prevent people from connecting to the network just because they can get to the parking lot.
Yep. My network doesn't seem to leave my yard, but there there may be more sensitive 802.11 cards. I have dealt with it by having my dhcp server send me mail if it sees a new MAC address. If I ever get that mail, I can decide what to do about it.
I'm making the assumption that an "attacker" (or freeloader) would try the simple thing of just hooking up and trying DHCP.
I may be out of touch with reality. Of all my friends with a ethernet network at home, they all also have a wireless network. Many of them have a wired network in only one or two rooms while the wireless covers the bulk of their houses (and part of the yards).
I know of a much smaller number of people with no wired network who are thinking of doing a wireless network.
My theory is that a wired network in one room is cheap and easy (the hardest part is getting an ISP to support it, or setting up NATing yourself). A wired network covering more then one room is normally kind of hard and/or costly (it is pretty easy and cheap if the house is still being built though!). A wireless network for a whole house is modestly expensive, but just as easy to install (plug a wireless card into all machines, plug in a access point).
So anyone with a single room network and get a whole house network quite easily by setting up 802.11. If you have a laptop it's a really easy choice :-) If you already have a multiroom network, I'm going to assume you are a true geek, and be quite intrested in having a wireless network as well. So both those cases make sense.
The no wired network, wireless only makes less sense. In part because it is so easy to do a one room network, and it is faster and somewhat safer. The one case where it makes sense is someone who can get a "cable modem" (or regular modem) into a room they don't want the computer in, and it is hard to get that access to the computer room. Then a wireless network might make more sense. I'm not sure people that don't want computers in every room really deserve wireless networks though :-)
Well I am picking at straws, it is my vocation, and my great skill. Oh, and I like it.
However if you are talking the 802.11 A&E, it doesn't seem to be so secure. Have you read slashdot recently? :-) Generically, yes, A&E solves all problems like that, but which A&E method? KerbV? SSH? SSL?
I think most 802.11 networks will be set up with no A&E, or with 802.11's WEP, which hasn't been all that successful. Those are the easy choices. Doing something else will be hard, nonstandard, and stand some chance of working.
(of corse easy and standard would be making the wireless network outside the normal firewall, but many places find that too inconvenient)
I don't have an ethernet printer at home, but I do use SSH tunnels there (to get my mail, and run VNC, and/or X).
The last place I worked had an open 802.11 network, but it was treated as "outside the firewall" by everything at work. They also have another open 802.11 network in another building that is inside the firewall, which is a bad idea.
I think the point is if you have a postscript printer on your IP network, and you hook up an 802.11 access point t your IP network, then your printer magically becomes an 802.11 printer as well (unless you fire wall off the 802.11 part, which is frequently a good idea, but not as frequently done).
The upside? One 802.11 access point gets all your printers. The downside? The longer range of 802.11 lets people outside your office waste your paper. Plus IP stacks on many printers are insanely fragile, so someone could break the printer pretty simply.