After looking through the basics, isn't the FileID something similar Hans did in Reiser?
Any filesystem that supports hard links has to have something like a Mac FileID. Traditional Unix filesystems call them "inode numbers" or "i-numbers", try "ls -i" sometime.
On Mac OS you can open a file given it's FileID but on most Unixes you can not (I assume OSX can, and some versions of SunOS/Solaris can as pat of one of the backup products). It opens a small security hole where you put files in a directory that denies permission to people but the files themselves allow it and you didn't want them to have permission (or only wanted them to have permission if they went through a specific program that knew how to make the filename). Mac OS can also map the FileID back into a name which most Unixes can't (Plan 9 can, and claims it is generally a great benefit, and far far simpler then they thought it would be back when the Plan9 folks invented Unix and decided it would be too hard to write and not worth it).
In short Unix has something exactly like FileIDs, but you can't use them in all the same ways despite the exact equivalence at the filesystem level.
For any other access pattern fragmentation doesn't matter as much, since you'll be skipping around the disk regardless of how the files are arranged.
For an extent based FS a less fragmented file is more likely to have whatever part of it's extent table is needed in memory to do a read or write on it. (for a block mapped one that is a non issue because the block map takes space proportional to the file size, not the number of fragments)
I'm not sure what your point about TRIPS is, but given that IBM themselves built the G5, any comparison between TRIPS & the G5 would surely have to take into account the same company assembled both...so there's an equality there that still makes benchmarks valid.
IBM fabs a bunch of things for other componies, and they don't get the same fab lines or processes (say no copper, or no SOI, or...). In fact even things they fab for their own reasons don't get fab'ed with the same things. For example (from another company) Intel uses it's best fabs on the x86 (and maybe the Itanic), but the fabs for the XSCALE (ARM) are a few gens old. The x86 needs the advanced fab to stay on the top of the pile, the XSCALE is in a less performance competitave market, but it is very price sensitave so it uses a fab that was state of the art during the P-II era and payed for itself by churning P-IIs out by the bucket load (I don't know that that is the actual fab, but it is the general idea).
Why is that important? Well it means just because IBM fabed both the G5 and TRIPS you can't use that as an "equality that makes the benchmarks valid"
How you ask? Simple. They've monopolized access to 1.8 inch hard drives. The other manufactureres of mp3 players aren't still using 2.5 inch drives because they're idiots and think consumers want a big clunky mp3 player, but because they simply can't get thier hands on anything else.
The question is how did they monopolize access to the 1.8 inch drives? If they merely told Toshiba "we'll buy as many as you can make at $X per" and nobody else wanted to pay that much (or more), then that ain't the least bit unfair.
If they did it by ending up being the only one that jumped when Toshiba said "we can make 1.8 inch drives, but it looks like nobody will buy them, so we aren't gonna quite finish figuring it out unless someone will commit to buying X of them" (and X being a number that hasn't been reached) then that ain't unfair either.
If they did it by standing up when Toshiba said "we can make 1.8 inch drives, who wants them?" and saying "we'll buy them for $X each, but you have to sell only to us for 3 years" then that is unfair on Apple's part and stupid on Toshiba's (note the diffrence between the last case and this one is mostly just that Toshiba is assumed to have the product ready to market here, and in the last case it was "we don't see the market to finish the R&D, find us a market and we'll reward your leap of faith with exclusave access" vs. "we have this product, so let's ignore an open market and just sell to Apple")
If they did it by telling Toshiba "sell to dell and we'll go elsewhere" then yeah, that would be unfair (and Toshiba beign the only maker of the drives would have to have been stupid to beleve them).
The thing is, how did they really get the strangle hold on the things? Somehow I susspect it is really that they were the only ones willing to pay what Toshiba wanted for them.
When music industry hypes up Brittany spears or Shakira (looks good, sounds bad) every geek worth his salt hates it. When Apple does the same thing its a good thing ?
Who says Apple has done the same thing? You have "Looks good, sounds bad" (Brittany/Shakira) on one hand and "Looks good, isn't perfect" (iPod) on the other. There is a big differnece between "that thing sucks but lots of people buy into the hype" and "that thing is pretty dern good (but there are better) and yet people buy into the hype".
And that's even assuming the 4 ways you have chosen to compare the Dell and iPod are the only 4 things that matter! Or that you even did it right (which matters more, what Dell writes on paper for the StoN, or what you can hear with your ears?)
How much can it cost to implement an Ogg Vorbis decoder anyway?
Testing costs money and takes time. The vague possiability that you might get sued by someone who thinks Ogg Vorbis violates their IP (even if it doesn't). You need a bit of space for the code which could have held the breakout game or something.
In other words "a little bit, but not much, unless you get unlucky".
But what do you get for it? There isn't a lot of demand (I say that, and I want it!), it doesn't let you make your product cheaper, it doesn't let you use a popular music service that folks want, it doesn't really do much past letting you put "OGG Vorbis support" on the box.
It's a hard sell, even with it being just this side of free!
"It has a better UI!" is not valid, since the competitor in question has comparable UI and UI's are subjective by nature.
A UI may be subjective, but one can study them anyway. The best method is to point a video camera at the user, and another at the device, and to give the user a list of things to do with the device and then for the researcher to leave the room. The UI that makes the fewest people frustrated tends to be better, as does the UI that people seem to learn faster, and depending on the task the one that either a new or experienced user can use faster.
It is easy to say "A and B have comparable UIs", but is it true that A and B have UIs that are not merely comparable, but actually roughly as good?
That is a question one can find the answer to, just not trivially.
Seriously, it seems to me that some people think iPod is the best simply because it's made by Apple. Regardless of the fact that there are players out there with superior features.
I'm more of the opinion that people think the iPod is better because it was the first to "not suck", and "everyone else has one". It may or may not also be the case that it is currently the best. There isn't a whole lot of blind love of Apple (if there were wouldn't it have more then 2~5% of the desktop market?) Oh, it may also be because the the iPod has some advertising and few of the other MP3 players have any (or as much).
(in the intrest of full disclosure, I have an iPod. I bought it when the 5G ones were pretty new and a modest amount of research showed nothing better on the market at the time. If it were to need replacing I would not automatically replace it with another iPod, but since I happen to use Mac OSX there is a pretty big chance that there won't be another MP3 player that works as well on that platform as the iPod. Then again a lot of the things I like about the iPod are really things I like about iTunes, so maybe it would work out)
Well, no. People keep on saying "iPod is the best mp3-player in the world!". I merely ask that how can it be "best" if there are players with far superior feature-sets? "
Well there is the chance that the features other MP3 players have are not all that important (or whatever features player B has that the iPod lacks are not as important to people as the features that player B has but the iPod lacks). For example while I would like it if the iPod played OGG files I'm not going to get a different MP3 player just because it plays OGG files, esp. if it lacks something that I really like about the iPod (like maybe how fast it sync's). I don't have any specific complaints about iRiver's player because it wasn't on the market when I picked the iPod.
You claimed that the UI is a subjective feature, and I agree. However you kind of missed that pretty much everything about an MP3 player is subjective. Yes you can establish that the battery in A lasts longer then the one in B, but you can't really say that makes A better then B because maybe a given user thinks B's battery lasts long enough. Worse yet some other user might think A is worse because B's battery charges faster or in a more convenient manner to him.
Unless product A is better in every way then product B you need to take into account that different people value different features differently and that while product B may be the clear winner to you someone else may decide that for them product A is far better. And you know what? You might both be right.
Should it say "just like Proteron's Lite Switch X...", or "just like Microsoft Windows."?
Well they gave MS credit for beign the first to do "Fast User Switching" (of corse they improved it enough to bash them a little at the same time, and I don't see 10.3's Cmd-Tab as any better then MS's Alt-Tab...then agian I dont' really find it better then 10.2's Cmd-Tab)...so sure, credit where credit is do due and all.
Are you telling me that Apple (and everybody else) aren't allowed to add new features to their products?
Did we read the same letter? I don't recall it telling Apple they shouldn't have added the feature, it in fact was positave about that. All it said was "give us credit! tell people we did it first!"
If the letter is to be beleved all they want is on the page that lists "150+ new features in panther" where they say "command tab now shows icons across the middle of the screen because we found out people like it that way" to say "just like Proteron's Lite Switch X we put icons across the middle of the display because we thought it looked cool they way they did it", or some such similar drivel.
I donno if it is the world's best idea, but it seems fair and like it won't stifle innovation in the least (unless of corse this "pixel perfect immatation" was actually totally random luck, and the guys that did it had never seen the other task switcher...which seems a little unlikely)
Re:Digital Rebel...delibratly cheaped out
on
Digital 35mm SLRs?
·
· Score: 1
the digital Rebel uses the same control system as the 35mm Rebel...
Pretty much (but at least it doesn't put image files on your flash card back to front...)
Remember not everyon on slashdot knows what the control system on the 35mm Rebel is like! Heck even a lot of non-EOS 35mm film owners have no idea.
and the quality of the picture depends on the quality of the lens...
Yep, more so then with the film EOS cameras though because you have fairly large cropping factor.
how is this +5 informative, again?
Because this is slashdot, not photo.net or dpreview.com? Not everyone here is a photographer. Not everyone knows a lot about interchangable lens systems, or SLRs, or even digital cameras. Not even the people that are thinking of plonking down $800 to $1500 for one.
Besides saying "the Rebel doesn't support CF4 focus=* so it sucks" is both untrue (it doesn't suck because of that, it is just a little more limited) and usless on slashdot (a low percentage of readers here know what that means...even fewer if I just said "CF4 is missing")
Basic facts about open source software ("no, it isn't stealing to use the GIMP for free", "once Open Source software supports your printer it probbalby always will, unlike MacOS or Windows where after enough upgrades support seems to vanish...and ok, if free software supports it OSX probbably will forever too, but don't let them off the hook for dropping color image writer support from OS9 to OSX!") would be thought of as insightful and informatave on photo.net and dpreview.
Does it really deserve +5? I donno, I guess if a few people buy whatever camera really suits them because of it, then yeah, I guess so. At least as much as half the other +5 posts on slashdot.
Re:Digital Rebel...delibratly cheaped out
on
Digital 35mm SLRs?
·
· Score: 1
ctually the EOS 50mm f/1.8 can be had for something like $65 *new* from places like BHphoto.com. On the DR, it becomes more like an 80-85mm, but I love it for handheld portraits.
I didn't really recall what I payed for my 50mm, but I'm pretty sure it was more then $65 from some local brick and mortor store. For $65 I really would urge anyone with a Canon SLR that doesn't already have a prime lens to go get the 50mm f/1.8.
Re:Digital Rebel...delibratly cheaped out
on
Digital 35mm SLRs?
·
· Score: 1
They had to cheapen it. It's not a pro camera.
The D10 isn't a pro camera either -- the focus screen isn't all that useable for manual focus and it isn't interchangable for ones that are. Not that you can't do pro work with any of those cameras, or in fact with any camera.
Who would buy the D10 if the D300 had all of the D10 features for $400 less?
Not many people, but I do want to be sure people don't buy the Digital Rebel without knowing what the extra $300 for the D10 (or the extra $2000 for the D1) gets them. It might be worth it to them, or it might not.
I love our D300.
Cool, I'm glad you like it. I'm gonna stick with my D30 for a bit longer, hopefully Canon will do a Digital ELAN with the EF-S mount before I decide to get another camera:-)
Digital Rebel...delibratly cheaped out
on
Digital 35mm SLRs?
·
· Score: 4, Informative
The image quality of prety much all the digital SLRs is very nice. Including the Digital Rebel. The focus time and shutter lag compaired ot the non SLR digitals is also very good (I have the now very old Canon D30, and while it has more shutter lag then the current digital SLRs it is low enough to get pictures of flying birds, or jumping dogs which I found really hard to do with compact digital cameras).
The digital rebel however suffers from being inteonally cheapened. It still takes great pictures, but if you had intended to use the camera in "manual mode" where you control both the shutter time and the apeature you'll find Canon decided to only put one dial on the camera. You have to switch between the two controls with a small button (there is also no way to assign auto focus to a button other then the shutter button). That's a royal pain if you ever get to a situation where you are smarter then the camera's light meter (and you'll run into them, digital cameras have less exposure latatude then print film, think of them more like slide film).
It also has cuppled the exposure mode and auto focus mode with the shooting mode. They took about 4 things that their other cameras let you set independantly and merged them into one thing and gave you maybe 12 choices, so a bunch of the combinations are not possiable.
Basically if your film SLR is a rebel you won't feel constrained by the digital rebal. If your film camera is an Elan you will be frustrated. If your digital camera is the point and click kind, then you will either be delighted or confused. Or both.
P.S. remember the camera is only the start of the spending:-) Lenses are very important. In fact the Digital Rebel's imager is better then most lenses. If you buy the DR and slap a $400 75-300mm USM-IS f/5.6 lens on it you won't get pictures nearly as sharp as the 300L f/4 lens...unfortuantly that lens costs quite a bit more then the camera. I strongly recomend at least one fast fixed focal length lens, the 50mm f/1.8 is in expsnave (under $100 used I think). It will show you how sharp your pictures can be, and more importantly it will let you get some natrual light shots where most zooms can't.
It is possible that there will never be a better type of "battery" than a hydrogen fuel cell, but can we be sure? The current vehicle fuel supply system has been around for a hundred years and cost fantastically large amounts of money. Are we so sure about hydrogen that we are willing to make a similar 100-year bet?
I don't think building the hydrogen distribution system is the big stumbling block. It is possiabble that we will just crack water with conventional eletric power to fill fuel cells, and that would just use the existing deployed eletric grid. The stumbling block is despite the large number of breakthroughs they just aren't cheap enough yet. Yes they are more efficent then other batteries (in terms of weight, as well as power loss), but they are just too costly. Once (if!) that gets fixed then they will show up. Maybe not in cars, but surely in laptops.
Besides if you go with a Fuel Cell system for cars and someone makes the "Super NiMH battery" that is better then Fuel Cells in every way you can rip the Fuel Cell out of the car and plop in the Super NiMH battery, after all they will both power the same eletric engine!
Besides we have been looking for a better battery for a long time, and in the last 30+ years (I know fuel cells are that old - they were part of the Apollo systems) we havn't found one. We might find a better one tomorrow, but I wouldn't bet on it.
don't even think we will avoid having batteries in our cars. In city driving, you can't beat the efficiency gains you can have from regenerative breaking. If your car needs a battery anyway, why not just go with a fully electric system?
Well 50 pounds of fuel cell plus 20 pounds of NiMH battery has got to take less energy to haul around then 3000 pounds of NiMH battery.
Why do you keep talking about a Fuel Cell system as if it is less eletric then a system that keeps it's power stored in a Lead Acid battery?
Why do we assume that electric cars have too short a range?
Because when GM attempted to sell an eletric car (the EV1) the lack of range was one of the very big complaints (along with lack of cargo capicity, lack of places to recharge, time to recharge, and price -- even though GM leased the car as if it cost $20,000 or so when it cost them more like $60,000 or so to make!).
So it isn't so much an asumption that eletronic cars have too short a range as an emperically gained fact. Now you can argue that it is a perception problem, that the range would have actually been acceptable buy people were too dumb to realise it. You might even be correct, but that won't make them part with their hard warned cash for a car that can only drive 60 miles on a charge (or whatever it was -- I seem to recall it was less then my daily round trip).
If we accept a shorter time between refueling, battery-powered cars might be the best way to go.
Sure, except people seem unwilling to accept it. It's not like GM didn't try.
Re:No difference for a long while, but...
on
The End of the Oil Age
·
· Score: 2, Insightful
Why add the conversion losses of splitting water and then recombining it again? Why not just use the electricity directly in the car?
There are three ways to "use electricity directly in the car":
Keep the car plugged into the mains while you drive -- which limits the car's range to the distanc eyou can drag the power coord.
Generate the power in the car -- which ties you to whatever you were using to gennerate it (petrol in things like the Honda Hybred and Toyta Prius)
Store it somehow and take it with you
Currently "store it and take it with you" is best done with a NiMH, NiCad, Lead-Acid, or some other kind of battery. In a eletric car like the GM EV1 well over half the weight of the car was battries. Battries that take a long time to charge, and wear out in just a few years (and cost more then many people were willing to pay for the car!).
Hydrogen Fuel Cells are merely another form of "store it and take it with you", but we think we can store way more power per pound, and that the fuel cells will last much longer and even be more efficent then current battries. Oh, and recharge much faster.
But I think the important part is to use electricity to power cars, this is the important part; not that electric power can be transferred from generators to individual vehicles using hydrogen as the transfer agent; that is surely less versatile than plugging in your car and charging up batteries with electricity right from the generator.
It is no more or less versatile since it is pretty much just another kind of battery. However it is hoped that it will be fantastically lighter, more efficent, faster to charge, and cheaper then existing battrey systems.
It may not be technically correct to think of Hydrogen Fuel Cells as a "much better battrey", but it is a very useful mental shortcut.
But people "whining" that their beige G3 or 3rd. party processor-upgraded pre-G3 PowerMac won't run Panther? I'm caught in the middle. Yes, they should probably accept the fact they've got to "pay to play" and upgrade their hardware. But no, I can't say I don't understand their pain, either. Computers are one of the worst "investments" around, as far as their resale value goes. New OS releases are usually an unpleasant reminder of why....
New comercial OS releases at least. Open source OSes do sometimes "de-support" old hardware, but normally only when the hardware is so old that nobody that ran beta versions of the hardware were avilable to test it! I have an ISA sound card bought in 1991 or so that worked under FreeBSD until about 2001. If I had really wanted it to keep working I could have spent a day or two figuring out what bit rot killed the driver and then submitted a patch and made the 3 other people on earth who still owned that card happy. I ended up spending $18 on a replacment, and leaving the work to on of those other 3 folks.
In fact I know someone who has been dilligantly working on adding MCA support to NetBSD (I think). He has been going to HAMfests and buying any MCA gear he hadn't seen before so he can code up a driver. Odd hobby, but I'm sure someone somewhere is thankful.
So why doesn't the beige G3 run Panther? Well I understand that Apple probbably doesn't see enough reason to assign testers to make sure it works, and it would probbabl ytake far more programmer time to fix the bugs then it did to code up the rotating cube while switchign users eye candy. Yeah, I understand that. But...
...do you really think once it is released the same folks that made Darwin boot on unsupported-by-Apple-old-Macs won't do their magic again and make the beige G3 work?
UNIX and X11 from 15 years ago is still source compatible with today's systems, and X11 is even fully binary compatible, not in emulation but natively.
I'm not sure X11 is 15 years old, it may well be only 14...
15 years ago Unix software largely assumed the return from read(2) was an int or long, not a ssize_t (and ssize_t might be a long long on modern systems and overflow an int). In fact 15 years ago a lot of software still suffered from "All the world is a VAX" sindrome, and was still being upgraded to the "All the world is a Sun" (which was a breif bump in the road to the enlightened modern "All the world is a x86 Linux"). You will find some Unix code from the late 80s plays fast and loose with things like struct passing. Yes, a struct of 4 ints can be passed to a function that expects 4 ints (and not a struct) as an argument. Oh, and some systems were so primitave that you could open a directory and call read(2) on it! I mean, imagine! (heck, some systems of the era required it!)
Oh, and don't make me tell you about when uids went from 8 bits to 16! that was far worse then the change from 32 to 64! (oh, has that happened yet?)
Lots of stuff from 1988 era Unix code isn't compatable with the modern stuff. Don't fool yourself, Unix code is pretty portable, but that doesn't mean anywhere close to 100% of "dusty deck C code" will just work.
Try to port University Ingress (which worked shoehorned into 64K split I and D on a PDP-11) to a "modern" Unix some day. Or wait 20 years and port one of today's programs to whatever is popular in the Unix world that year.
Personally, I think its really bad, and I feel bad for the users, but Apple has historically not really cared all *that* much about the end users. If it cant go on the JumboTron behind an AppleWorld keynote speach, it isnt really that important to them.
Maybe, maybe not. The kind of things Apple use to break compatability with were "if your program reaches into the finder and looks at the undocumented gPbtrXyzzy variable, well we made changes to the finder, and not only has gPbtrXyzzy moved, it doesn't mean exactly the same thing anymore, buy hey, it lets the user copy more then one file at a time, so good for them!".
If that general level of thing is what the continue to do, then it is good in the long run. It lets them change stuff that needs changing, and it reminds Mac coders to stick to documented interfaces as much as they possiabbably can.
My Powerbook has Bluetooth. The thing is, not one device other than my Powerbook has it. The only device I'm even thinking of getting that could have it is a new cellphone.
The only two things I have with BT are a cell phone and a headset (and oddly enough the Mac doesn't support the headset, so I can't use it as a speaker or mic with the Mac). Then again I don't have many other things I would want BT on. Maybe my camera, but I would rather have something faster on that (I don't use it's USB cable, so I wouldn't want "wireless USB " for it)
I'd like to have some sort of connection between the laptop and phone for doing internet connectivity and, perhaps most importantly, managing the phone's contact list. But a USB wire would do the trick just fine, as it does for my Palm.
I have a Palm that sync's over USB, and the phone that sync's over BT. Guess which gets sync'ed far more frequently? It is also nice since I have my laptop configured to use the phone's GPRS connection when it can't find 802.11 or a wired connection. Nice not to have to hook it up. Not the end-all or anything, but very nice.
And, since it interferes with 802.11 which I use nearly 100% of the time, I'm certainly not motivated to use it for anything other than intermittent device communication.
It doesn't seem to, I mean it must in theory since they share a frequency band, but it doesn't seem to drop my Tx speed, or the signal strength bar, nor when I've got no signal does turning BT off get me some. The effect must be pretty marginal.
I think I'd rather just have 802.11 devices, seeing as those are getting pretty ubiquitous.
Me too, and I would rather have FireWire or 100Mb (or 1Gb) ethernet then USB, but does that really make a lot of sense for hooking up a mouse?
What I would like to see is an evolution of 802.11 client software such that they can self-organize with checks. For instance, I would really like to be able to have my AP notify me that a new client wants to join, identify the device as best possible, and allow me to say "yes" or "no." If I approve it, I'd like the client to be configured properly by the server
I have my 802.11 access point configured not to act as a DHCP server, but to pass the DHCP on to my FreeBSD box. I have it set up to send me mail about MAC address it hasn't heard of before (and grant them and address -- but it is trivial to make it deny by default). The same dhcp server (ISC's DHCP server) is avilable for Linux and other Unix & Unix like systems. Most 802.11 access points can even be configured to deny the associate request baised off of a RADIUS query whoch would be slightly better (it would make it harder for someone you don't want connecting to sniff the traffic and steal an IP address), but not by enough that I have bothered to do anything about it in the last 3 years.
In this way configuration would be nothing more complicated than picking the SIID from a drop down or typing one in.
That'll be great for palm's and cell phones, but how would it work with a headset (audio I/O, one single color LED, and maybe 3 buttons), a mouse (lots of input, but no display), or a keyboard (again no display)? BT gets around that by being able to drive the selection mostly from either side (hold the one button down for 10 seconds, then drive the process from the computer or cellphone).
I would like those devices to be 802.11, fewer things to muck with, and simpler for me to write drivers for (and it would work for EV-DO as well as GPRS), but it seems like it would be difficult to make work in all cases.
OSS (not necessarily Linux) is indeed the best solution in all cases in which 1) you need an OS and 2) need only an OS and no special apps/drivers/niche-features/etc (because you create all your apps yourself).
A few more places where CSS might beat OSS.
the minimum hardware needed to run your app doesn't run an OSS OS, like if all you need is a 6502 at 4Mhz it makes little sense to buy a 386SX so you can run Linux (of corse int he low end case you frequently write the OS on your own, you need to make sure you will ship enough units to that the $10 cheaper hardware will cover the extra NRE of working in such constrained hardware and maybe writing the OS as well)
The CSS might perform better (I guess you could call that a "niche-feature" though), for example QNX services interrupts faster then other x86 OSes, or at least it did last I checked.
The CSS's vender might be attempting to break into a market and not only match OSS's $0 per unit price, but throw in some resources like marketing money and maybe programmers or something to help sweeten the deal.
The CSS has more people that you can hire at lower wages to work on it (maybe more people have MS Access on their resume then MySQL, and they work for less...of corse in that case you may end up needing far more of them for far longer to make a less stable product, but hey it looks good on paper)
Your company makes CSS OSes that can run on the hardware you are using. I mean would MS ship a wrist watch running Linux even if it was in all ways better then "WinCE for wrist tops"?
That's why 5 years ago about 2/3 of embedded systems with OSes used in-house OSes, while now the majority of embedded projects already use Linux.
Does that include CoinOp video games? (that was the last place I did embedded work...er, except for that other thing)
I got the impression that the goal was to find a way to replace specialized equipment that costs hundreds of thousands, if not millions of dollars, with equipment that was much cheaper.
I think you are right...however that gives a free OS like Linux a few advantages. One there is no purchase cost for the OS, so your per unit costs go down (by $200 if you have to buy WinXP retail, or maybe by as little as $5 if you buy a huge number of WinCE licencess). Two you can run on a wider array of hardware then most other OSes, so you might be able to save money on the hardware. For "any comercial OS" as opposed to "that one comercial OS" this isn't really an advantage since a given set of hardware tends to have at least one non-free OS that runs on it. So if you want to run on a FROON2000 CPU and pay money for QNX you probbably can, if you want to run WinXP on it then the FROON2000 had better really be an x86 of some sort.
On the plus side sometimes comercial OSes have drivers for stuff free OSes don't, or are faster. QNX does context switches faster then Linux (or did last I checked), and might let you get away with slower (and cheaper!) hardware for the same job, and if that price reduction is more then QNX's cost then it is a win. Simillarally Windows can drive lots of softmodems and Linux can drive very few, if your productr needed a modem, and the cheapest one was a softmodem that Linux didn't drive that could be an issue (not likely because modems cost a lot less then Windows, so unless your project has to deal with hardware that is already deployed...).
The whole 'the provider will screw you' argument is dubious. Windows and Mac OS X boxen are commodities, so its not as if Apple or MS will be able to apply pressure to some community using store bought machines to give them more money.
Well if they had decided to use OSX then they would have to pay Apple's hardware prices, which are not all that cheap. They would also be stuck with a hardware platform that seems to change once a year or two. If they go with Windows they might be stuck with some product regrstration BS that gets in the way of the turnkey operation (WinCE probbably wouldn't have that problem though). Comercial OSes do get in the way sometimes.
My point isn't that linux should not have been used. It was most likely the best choice and had I been a part of the project I would have supported Linux over any other OS. It can be argued that it wasn't the only possible choice for reducing the cost of such systems by an order of magnitude.
You are right, use of Linux probbably "only" saved a few $100 per unit, and since it looks like they are already saving $10,000 that last $100 isn't a big deal.
Would I have chosen Linux? Well, not really. I'm more of a BSD fan, but the same arguments apply there, and Linux was probbably as good a choice as any of the BSDs (except maybe OpenBSD...or in a few years Dragonfly).
Bloody hell! Are you guys in the US still running analog mobile services?
Yes, in the sort of areas that have similar population densities to Scotland's non-served cell areas (and to be fair, some parts of Scotland that get lots of tourists but still have low population density do have digital cell coverage while in the USA areas there are few areas like that...there are also places in Scotland that have digital coverage while the equivalent part of the USA has analog only, but on the whole I would say there isn't too much of that)
The old brown stuff looks like the kit I was ripping out back in 1997 - clunky old analog cell switches
I would hazard a guess that the analog kit here is older then 1997, most cell phone build-outs post 1990 here were digital, or at the very least DAMPS, but there were a lot of cell phone build-outs prior to 1990. Cell phones have been with us for a very long time (I seem to recall some people putting the start of the US cell stuff at 1952 or so, but I think those were really radio phones), and only the high use areas or areas with new competitors since about 1990 have seen new build-outs. The other areas "suffer" with whatever they were first installed with until it becomes cheaper to upgrade it then to fix it.
You're describing something that is technically impossible. It just can't happen. I'd suggest you check your bandwidth use as you're most certainly connecting to a network outside your own with a dedicated base station, probably a neighbour above or below you.
It's not impossible, in this case the Mac laptop (or was it a desktop?) is hooked up to the cable modem via an ethernet. Then one of two things can be done:
Have that Mac start an ad hoc network
Have that Mac act as the 802.11 Access Point (access points don't need different hardware, many just have normal PCMCIA cards in them! It is all just software, and nothing prevents a normal desktop or server OS from having such software!)
I have a sneaking suspicion that offering to share the network (through the GUI) makes an ad hoc network. So you will use slightly more power (infrastructure mode lets the access point tell the other devices when they need to listen, so they can be powered down part of the time and still get all their traffic). For the most part that won't make any difference.
As far as the hardware goes, there is nothing that should be able to prevent one 802.11 card from being on multiple 802.11 networks at once, but I don't know of any software that attempts to do it, and given how the cards operate they might all have to be on the same channel. As far as hardware and software go there should be nothing preventing one host from having more then one 802.11 card and either routing across them, or bridging. In fact I have done that.
OpenSSH people have developed their product for BSD and Linux so checking the update takes less time. of OS X Apple probably needs to tweak it a little.
Last time there was a ssh security bug I fetched down the "portable ssh" and it just compiled on OSX. Zero tweaking. That's not supprising because at that level it looks exactly like all the other BSDs, same calls to get psudottys, same setuid calls and so on.
I'm guessing it takes Apple a lot of time because they have to put it in a pretty package, and test it on 37 different machines. I only test it on one (hey that is 50% of the Mac I'm the admin on!).
then it would be hard to claim that the later rip just happened to have the same errors - assuming errors are essentially random.
But are they? Or are they tied to defects on the CD (scratches on the same place, or a bit error in the glass master?) Or are they tied to the model of CD-ROM used to read them? Or just plain random?
If it is tied to the CD-ROM one could argue that you ripped that CD with the same model CD-ROM as that other guy over there, but when the CD-ROM broke you got a new one (and it could even be true). (this isn't a claim I'll be needign to make, I actually ripped all my own MP3's, it's hard to find a good rip out there...and besides I'm on dial-up)
Slashdot Mirror
Any filesystem that supports hard links has to have something like a Mac FileID. Traditional Unix filesystems call them "inode numbers" or "i-numbers", try "ls -i" sometime.
On Mac OS you can open a file given it's FileID but on most Unixes you can not (I assume OSX can, and some versions of SunOS/Solaris can as pat of one of the backup products). It opens a small security hole where you put files in a directory that denies permission to people but the files themselves allow it and you didn't want them to have permission (or only wanted them to have permission if they went through a specific program that knew how to make the filename). Mac OS can also map the FileID back into a name which most Unixes can't (Plan 9 can, and claims it is generally a great benefit, and far far simpler then they thought it would be back when the Plan9 folks invented Unix and decided it would be too hard to write and not worth it).
In short Unix has something exactly like FileIDs, but you can't use them in all the same ways despite the exact equivalence at the filesystem level.
For an extent based FS a less fragmented file is more likely to have whatever part of it's extent table is needed in memory to do a read or write on it. (for a block mapped one that is a non issue because the block map takes space proportional to the file size, not the number of fragments)
IBM fabs a bunch of things for other componies, and they don't get the same fab lines or processes (say no copper, or no SOI, or...). In fact even things they fab for their own reasons don't get fab'ed with the same things. For example (from another company) Intel uses it's best fabs on the x86 (and maybe the Itanic), but the fabs for the XSCALE (ARM) are a few gens old. The x86 needs the advanced fab to stay on the top of the pile, the XSCALE is in a less performance competitave market, but it is very price sensitave so it uses a fab that was state of the art during the P-II era and payed for itself by churning P-IIs out by the bucket load (I don't know that that is the actual fab, but it is the general idea).
Why is that important? Well it means just because IBM fabed both the G5 and TRIPS you can't use that as an "equality that makes the benchmarks valid"
The question is how did they monopolize access to the 1.8 inch drives? If they merely told Toshiba "we'll buy as many as you can make at $X per" and nobody else wanted to pay that much (or more), then that ain't the least bit unfair.
If they did it by ending up being the only one that jumped when Toshiba said "we can make 1.8 inch drives, but it looks like nobody will buy them, so we aren't gonna quite finish figuring it out unless someone will commit to buying X of them" (and X being a number that hasn't been reached) then that ain't unfair either.
If they did it by standing up when Toshiba said "we can make 1.8 inch drives, who wants them?" and saying "we'll buy them for $X each, but you have to sell only to us for 3 years" then that is unfair on Apple's part and stupid on Toshiba's (note the diffrence between the last case and this one is mostly just that Toshiba is assumed to have the product ready to market here, and in the last case it was "we don't see the market to finish the R&D, find us a market and we'll reward your leap of faith with exclusave access" vs. "we have this product, so let's ignore an open market and just sell to Apple")
If they did it by telling Toshiba "sell to dell and we'll go elsewhere" then yeah, that would be unfair (and Toshiba beign the only maker of the drives would have to have been stupid to beleve them).
The thing is, how did they really get the strangle hold on the things? Somehow I susspect it is really that they were the only ones willing to pay what Toshiba wanted for them.
Who says Apple has done the same thing? You have "Looks good, sounds bad" (Brittany/Shakira) on one hand and "Looks good, isn't perfect" (iPod) on the other. There is a big differnece between "that thing sucks but lots of people buy into the hype" and "that thing is pretty dern good (but there are better) and yet people buy into the hype".
And that's even assuming the 4 ways you have chosen to compare the Dell and iPod are the only 4 things that matter! Or that you even did it right (which matters more, what Dell writes on paper for the StoN, or what you can hear with your ears?)
Testing costs money and takes time. The vague possiability that you might get sued by someone who thinks Ogg Vorbis violates their IP (even if it doesn't). You need a bit of space for the code which could have held the breakout game or something.
In other words "a little bit, but not much, unless you get unlucky".
But what do you get for it? There isn't a lot of demand (I say that, and I want it!), it doesn't let you make your product cheaper, it doesn't let you use a popular music service that folks want, it doesn't really do much past letting you put "OGG Vorbis support" on the box.
It's a hard sell, even with it being just this side of free!
A UI may be subjective, but one can study them anyway. The best method is to point a video camera at the user, and another at the device, and to give the user a list of things to do with the device and then for the researcher to leave the room. The UI that makes the fewest people frustrated tends to be better, as does the UI that people seem to learn faster, and depending on the task the one that either a new or experienced user can use faster.
It is easy to say "A and B have comparable UIs", but is it true that A and B have UIs that are not merely comparable, but actually roughly as good?
That is a question one can find the answer to, just not trivially.
I'm more of the opinion that people think the iPod is better because it was the first to "not suck", and "everyone else has one". It may or may not also be the case that it is currently the best. There isn't a whole lot of blind love of Apple (if there were wouldn't it have more then 2~5% of the desktop market?) Oh, it may also be because the the iPod has some advertising and few of the other MP3 players have any (or as much).
(in the intrest of full disclosure, I have an iPod. I bought it when the 5G ones were pretty new and a modest amount of research showed nothing better on the market at the time. If it were to need replacing I would not automatically replace it with another iPod, but since I happen to use Mac OSX there is a pretty big chance that there won't be another MP3 player that works as well on that platform as the iPod. Then again a lot of the things I like about the iPod are really things I like about iTunes, so maybe it would work out)
Well there is the chance that the features other MP3 players have are not all that important (or whatever features player B has that the iPod lacks are not as important to people as the features that player B has but the iPod lacks). For example while I would like it if the iPod played OGG files I'm not going to get a different MP3 player just because it plays OGG files, esp. if it lacks something that I really like about the iPod (like maybe how fast it sync's). I don't have any specific complaints about iRiver's player because it wasn't on the market when I picked the iPod.
You claimed that the UI is a subjective feature, and I agree. However you kind of missed that pretty much everything about an MP3 player is subjective. Yes you can establish that the battery in A lasts longer then the one in B, but you can't really say that makes A better then B because maybe a given user thinks B's battery lasts long enough. Worse yet some other user might think A is worse because B's battery charges faster or in a more convenient manner to him.
Unless product A is better in every way then product B you need to take into account that different people value different features differently and that while product B may be the clear winner to you someone else may decide that for them product A is far better. And you know what? You might both be right.
It won't stop anyone form arguing though :-)
Well they gave MS credit for beign the first to do "Fast User Switching" (of corse they improved it enough to bash them a little at the same time, and I don't see 10.3's Cmd-Tab as any better then MS's Alt-Tab...then agian I dont' really find it better then 10.2's Cmd-Tab)...so sure, credit where credit is do due and all.
Did we read the same letter? I don't recall it telling Apple they shouldn't have added the feature, it in fact was positave about that. All it said was "give us credit! tell people we did it first!"
If the letter is to be beleved all they want is on the page that lists "150+ new features in panther" where they say "command tab now shows icons across the middle of the screen because we found out people like it that way" to say "just like Proteron's Lite Switch X we put icons across the middle of the display because we thought it looked cool they way they did it", or some such similar drivel.
I donno if it is the world's best idea, but it seems fair and like it won't stifle innovation in the least (unless of corse this "pixel perfect immatation" was actually totally random luck, and the guys that did it had never seen the other task switcher...which seems a little unlikely)
Pretty much (but at least it doesn't put image files on your flash card back to front...)
Remember not everyon on slashdot knows what the control system on the 35mm Rebel is like! Heck even a lot of non-EOS 35mm film owners have no idea.
Yep, more so then with the film EOS cameras though because you have fairly large cropping factor.
Because this is slashdot, not photo.net or dpreview.com? Not everyone here is a photographer. Not everyone knows a lot about interchangable lens systems, or SLRs, or even digital cameras. Not even the people that are thinking of plonking down $800 to $1500 for one.
Besides saying "the Rebel doesn't support CF4 focus=* so it sucks" is both untrue (it doesn't suck because of that, it is just a little more limited) and usless on slashdot (a low percentage of readers here know what that means...even fewer if I just said "CF4 is missing")
Basic facts about open source software ("no, it isn't stealing to use the GIMP for free", "once Open Source software supports your printer it probbalby always will, unlike MacOS or Windows where after enough upgrades support seems to vanish...and ok, if free software supports it OSX probbably will forever too, but don't let them off the hook for dropping color image writer support from OS9 to OSX!") would be thought of as insightful and informatave on photo.net and dpreview.
Does it really deserve +5? I donno, I guess if a few people buy whatever camera really suits them because of it, then yeah, I guess so. At least as much as half the other +5 posts on slashdot.
I didn't really recall what I payed for my 50mm, but I'm pretty sure it was more then $65 from some local brick and mortor store. For $65 I really would urge anyone with a Canon SLR that doesn't already have a prime lens to go get the 50mm f/1.8.
The D10 isn't a pro camera either -- the focus screen isn't all that useable for manual focus and it isn't interchangable for ones that are. Not that you can't do pro work with any of those cameras, or in fact with any camera.
Not many people, but I do want to be sure people don't buy the Digital Rebel without knowing what the extra $300 for the D10 (or the extra $2000 for the D1) gets them. It might be worth it to them, or it might not.
Cool, I'm glad you like it. I'm gonna stick with my D30 for a bit longer, hopefully Canon will do a Digital ELAN with the EF-S mount before I decide to get another camera :-)
The image quality of prety much all the digital SLRs is very nice. Including the Digital Rebel. The focus time and shutter lag compaired ot the non SLR digitals is also very good (I have the now very old Canon D30, and while it has more shutter lag then the current digital SLRs it is low enough to get pictures of flying birds, or jumping dogs which I found really hard to do with compact digital cameras).
The digital rebel however suffers from being inteonally cheapened. It still takes great pictures, but if you had intended to use the camera in "manual mode" where you control both the shutter time and the apeature you'll find Canon decided to only put one dial on the camera. You have to switch between the two controls with a small button (there is also no way to assign auto focus to a button other then the shutter button). That's a royal pain if you ever get to a situation where you are smarter then the camera's light meter (and you'll run into them, digital cameras have less exposure latatude then print film, think of them more like slide film).
It also has cuppled the exposure mode and auto focus mode with the shooting mode. They took about 4 things that their other cameras let you set independantly and merged them into one thing and gave you maybe 12 choices, so a bunch of the combinations are not possiable.
Basically if your film SLR is a rebel you won't feel constrained by the digital rebal. If your film camera is an Elan you will be frustrated. If your digital camera is the point and click kind, then you will either be delighted or confused. Or both.
P.S. remember the camera is only the start of the spending :-) Lenses are very important. In fact the Digital Rebel's imager is better then most lenses. If you buy the DR and slap a $400 75-300mm USM-IS f/5.6 lens on it you won't get pictures nearly as sharp as the 300L f/4 lens...unfortuantly that lens costs quite a bit more then the camera. I strongly recomend at least one fast fixed focal length lens, the 50mm f/1.8 is in expsnave (under $100 used I think). It will show you how sharp your pictures can be, and more importantly it will let you get some natrual light shots where most zooms can't.
I don't think building the hydrogen distribution system is the big stumbling block. It is possiabble that we will just crack water with conventional eletric power to fill fuel cells, and that would just use the existing deployed eletric grid. The stumbling block is despite the large number of breakthroughs they just aren't cheap enough yet. Yes they are more efficent then other batteries (in terms of weight, as well as power loss), but they are just too costly. Once (if!) that gets fixed then they will show up. Maybe not in cars, but surely in laptops.
Besides if you go with a Fuel Cell system for cars and someone makes the "Super NiMH battery" that is better then Fuel Cells in every way you can rip the Fuel Cell out of the car and plop in the Super NiMH battery, after all they will both power the same eletric engine!
Besides we have been looking for a better battery for a long time, and in the last 30+ years (I know fuel cells are that old - they were part of the Apollo systems) we havn't found one. We might find a better one tomorrow, but I wouldn't bet on it.
Well 50 pounds of fuel cell plus 20 pounds of NiMH battery has got to take less energy to haul around then 3000 pounds of NiMH battery.
Why do you keep talking about a Fuel Cell system as if it is less eletric then a system that keeps it's power stored in a Lead Acid battery?
Because when GM attempted to sell an eletric car (the EV1) the lack of range was one of the very big complaints (along with lack of cargo capicity, lack of places to recharge, time to recharge, and price -- even though GM leased the car as if it cost $20,000 or so when it cost them more like $60,000 or so to make!).
So it isn't so much an asumption that eletronic cars have too short a range as an emperically gained fact. Now you can argue that it is a perception problem, that the range would have actually been acceptable buy people were too dumb to realise it. You might even be correct, but that won't make them part with their hard warned cash for a car that can only drive 60 miles on a charge (or whatever it was -- I seem to recall it was less then my daily round trip).
Sure, except people seem unwilling to accept it. It's not like GM didn't try.
There are three ways to "use electricity directly in the car":
Currently "store it and take it with you" is best done with a NiMH, NiCad, Lead-Acid, or some other kind of battery. In a eletric car like the GM EV1 well over half the weight of the car was battries. Battries that take a long time to charge, and wear out in just a few years (and cost more then many people were willing to pay for the car!).
Hydrogen Fuel Cells are merely another form of "store it and take it with you", but we think we can store way more power per pound, and that the fuel cells will last much longer and even be more efficent then current battries. Oh, and recharge much faster.
It is no more or less versatile since it is pretty much just another kind of battery. However it is hoped that it will be fantastically lighter, more efficent, faster to charge, and cheaper then existing battrey systems.
It may not be technically correct to think of Hydrogen Fuel Cells as a "much better battrey", but it is a very useful mental shortcut.
New comercial OS releases at least. Open source OSes do sometimes "de-support" old hardware, but normally only when the hardware is so old that nobody that ran beta versions of the hardware were avilable to test it! I have an ISA sound card bought in 1991 or so that worked under FreeBSD until about 2001. If I had really wanted it to keep working I could have spent a day or two figuring out what bit rot killed the driver and then submitted a patch and made the 3 other people on earth who still owned that card happy. I ended up spending $18 on a replacment, and leaving the work to on of those other 3 folks.
In fact I know someone who has been dilligantly working on adding MCA support to NetBSD (I think). He has been going to HAMfests and buying any MCA gear he hadn't seen before so he can code up a driver. Odd hobby, but I'm sure someone somewhere is thankful.
So why doesn't the beige G3 run Panther? Well I understand that Apple probbably doesn't see enough reason to assign testers to make sure it works, and it would probbabl ytake far more programmer time to fix the bugs then it did to code up the rotating cube while switchign users eye candy. Yeah, I understand that. But...
...do you really think once it is released the same folks that made Darwin boot on unsupported-by-Apple-old-Macs won't do their magic again and make the beige G3 work?
I betcha $5 it'll work by the end of November.
I'm not sure X11 is 15 years old, it may well be only 14...
15 years ago Unix software largely assumed the return from read(2) was an int or long, not a ssize_t (and ssize_t might be a long long on modern systems and overflow an int). In fact 15 years ago a lot of software still suffered from "All the world is a VAX" sindrome, and was still being upgraded to the "All the world is a Sun" (which was a breif bump in the road to the enlightened modern "All the world is a x86 Linux"). You will find some Unix code from the late 80s plays fast and loose with things like struct passing. Yes, a struct of 4 ints can be passed to a function that expects 4 ints (and not a struct) as an argument. Oh, and some systems were so primitave that you could open a directory and call read(2) on it! I mean, imagine! (heck, some systems of the era required it!)
Oh, and don't make me tell you about when uids went from 8 bits to 16! that was far worse then the change from 32 to 64! (oh, has that happened yet?)
Lots of stuff from 1988 era Unix code isn't compatable with the modern stuff. Don't fool yourself, Unix code is pretty portable, but that doesn't mean anywhere close to 100% of "dusty deck C code" will just work.
Try to port University Ingress (which worked shoehorned into 64K split I and D on a PDP-11) to a "modern" Unix some day. Or wait 20 years and port one of today's programs to whatever is popular in the Unix world that year.
Maybe, maybe not. The kind of things Apple use to break compatability with were "if your program reaches into the finder and looks at the undocumented gPbtrXyzzy variable, well we made changes to the finder, and not only has gPbtrXyzzy moved, it doesn't mean exactly the same thing anymore, buy hey, it lets the user copy more then one file at a time, so good for them!".
If that general level of thing is what the continue to do, then it is good in the long run. It lets them change stuff that needs changing, and it reminds Mac coders to stick to documented interfaces as much as they possiabbably can.
The only two things I have with BT are a cell phone and a headset (and oddly enough the Mac doesn't support the headset, so I can't use it as a speaker or mic with the Mac). Then again I don't have many other things I would want BT on. Maybe my camera, but I would rather have something faster on that (I don't use it's USB cable, so I wouldn't want "wireless USB " for it)
I have a Palm that sync's over USB, and the phone that sync's over BT. Guess which gets sync'ed far more frequently? It is also nice since I have my laptop configured to use the phone's GPRS connection when it can't find 802.11 or a wired connection. Nice not to have to hook it up. Not the end-all or anything, but very nice.
It doesn't seem to, I mean it must in theory since they share a frequency band, but it doesn't seem to drop my Tx speed, or the signal strength bar, nor when I've got no signal does turning BT off get me some. The effect must be pretty marginal.
Me too, and I would rather have FireWire or 100Mb (or 1Gb) ethernet then USB, but does that really make a lot of sense for hooking up a mouse?
I have my 802.11 access point configured not to act as a DHCP server, but to pass the DHCP on to my FreeBSD box. I have it set up to send me mail about MAC address it hasn't heard of before (and grant them and address -- but it is trivial to make it deny by default). The same dhcp server (ISC's DHCP server) is avilable for Linux and other Unix & Unix like systems. Most 802.11 access points can even be configured to deny the associate request baised off of a RADIUS query whoch would be slightly better (it would make it harder for someone you don't want connecting to sniff the traffic and steal an IP address), but not by enough that I have bothered to do anything about it in the last 3 years.
That'll be great for palm's and cell phones, but how would it work with a headset (audio I/O, one single color LED, and maybe 3 buttons), a mouse (lots of input, but no display), or a keyboard (again no display)? BT gets around that by being able to drive the selection mostly from either side (hold the one button down for 10 seconds, then drive the process from the computer or cellphone).
I would like those devices to be 802.11, fewer things to muck with, and simpler for me to write drivers for (and it would work for EV-DO as well as GPRS), but it seems like it would be difficult to make work in all cases.
A few more places where CSS might beat OSS.
Does that include CoinOp video games? (that was the last place I did embedded work...er, except for that other thing)
I think you are right...however that gives a free OS like Linux a few advantages. One there is no purchase cost for the OS, so your per unit costs go down (by $200 if you have to buy WinXP retail, or maybe by as little as $5 if you buy a huge number of WinCE licencess). Two you can run on a wider array of hardware then most other OSes, so you might be able to save money on the hardware. For "any comercial OS" as opposed to "that one comercial OS" this isn't really an advantage since a given set of hardware tends to have at least one non-free OS that runs on it. So if you want to run on a FROON2000 CPU and pay money for QNX you probbably can, if you want to run WinXP on it then the FROON2000 had better really be an x86 of some sort.
On the plus side sometimes comercial OSes have drivers for stuff free OSes don't, or are faster. QNX does context switches faster then Linux (or did last I checked), and might let you get away with slower (and cheaper!) hardware for the same job, and if that price reduction is more then QNX's cost then it is a win. Simillarally Windows can drive lots of softmodems and Linux can drive very few, if your productr needed a modem, and the cheapest one was a softmodem that Linux didn't drive that could be an issue (not likely because modems cost a lot less then Windows, so unless your project has to deal with hardware that is already deployed...).
Well if they had decided to use OSX then they would have to pay Apple's hardware prices, which are not all that cheap. They would also be stuck with a hardware platform that seems to change once a year or two. If they go with Windows they might be stuck with some product regrstration BS that gets in the way of the turnkey operation (WinCE probbably wouldn't have that problem though). Comercial OSes do get in the way sometimes.
You are right, use of Linux probbably "only" saved a few $100 per unit, and since it looks like they are already saving $10,000 that last $100 isn't a big deal.
Would I have chosen Linux? Well, not really. I'm more of a BSD fan, but the same arguments apply there, and Linux was probbably as good a choice as any of the BSDs (except maybe OpenBSD...or in a few years Dragonfly).
Yes, in the sort of areas that have similar population densities to Scotland's non-served cell areas (and to be fair, some parts of Scotland that get lots of tourists but still have low population density do have digital cell coverage while in the USA areas there are few areas like that...there are also places in Scotland that have digital coverage while the equivalent part of the USA has analog only, but on the whole I would say there isn't too much of that)
I would hazard a guess that the analog kit here is older then 1997, most cell phone build-outs post 1990 here were digital, or at the very least DAMPS, but there were a lot of cell phone build-outs prior to 1990. Cell phones have been with us for a very long time (I seem to recall some people putting the start of the US cell stuff at 1952 or so, but I think those were really radio phones), and only the high use areas or areas with new competitors since about 1990 have seen new build-outs. The other areas "suffer" with whatever they were first installed with until it becomes cheaper to upgrade it then to fix it.
It's not impossible, in this case the Mac laptop (or was it a desktop?) is hooked up to the cable modem via an ethernet. Then one of two things can be done:
I have a sneaking suspicion that offering to share the network (through the GUI) makes an ad hoc network. So you will use slightly more power (infrastructure mode lets the access point tell the other devices when they need to listen, so they can be powered down part of the time and still get all their traffic). For the most part that won't make any difference.
As far as the hardware goes, there is nothing that should be able to prevent one 802.11 card from being on multiple 802.11 networks at once, but I don't know of any software that attempts to do it, and given how the cards operate they might all have to be on the same channel. As far as hardware and software go there should be nothing preventing one host from having more then one 802.11 card and either routing across them, or bridging. In fact I have done that.
Last time there was a ssh security bug I fetched down the "portable ssh" and it just compiled on OSX. Zero tweaking. That's not supprising because at that level it looks exactly like all the other BSDs, same calls to get psudottys, same setuid calls and so on.
I'm guessing it takes Apple a lot of time because they have to put it in a pretty package, and test it on 37 different machines. I only test it on one (hey that is 50% of the Mac I'm the admin on!).
But are they? Or are they tied to defects on the CD (scratches on the same place, or a bit error in the glass master?) Or are they tied to the model of CD-ROM used to read them? Or just plain random?
If it is tied to the CD-ROM one could argue that you ripped that CD with the same model CD-ROM as that other guy over there, but when the CD-ROM broke you got a new one (and it could even be true). (this isn't a claim I'll be needign to make, I actually ripped all my own MP3's, it's hard to find a good rip out there...and besides I'm on dial-up)