I know this is going to be a flamebait. But before you flame me, consider the following: I'm researcher and get paid for what I do. I've released quite a few codes as open source and invented a bunch of algorithms which are not patented and used in many applications (think email spam filter, face recognition, etc.). And I've worked in industry and academia. For almost two decades. So I know both open and closed source.
First off, ideas have value. As in Dollar value. Take NVIDIA for instance - they don't have a semiconductor fab, so they send their chip layout to a place like TSMC or Global Foundry or Samsung or any other place to have their files turned into chips. These places are like modern printing presses. If their mask, vhdl or layout information were open source they wouldn't be able to reap the benefit from their investment into building the next generation of chips. Or as a more extreme case, take ARM. They design processor cores and license the microarchitecture to other (possibly fabless) design companies such as Apple which, in turn, tweak the design, add more stuff to it, and then ship it to the foundries. In other words, all the good stuff is in the plans, much less in the actual hardware.
So, designing an open source CPU is probably not going to work. Why not? Well, unlike with software, there's a massive barrier to entry. Talk Millions of Dollars rather than a few hundred to buy a laptop and install some version of GCC on it. Few users can afford this. This pretty much kills the model where many users take advantage of a good idea and share it to make it better. Yes, there are good ideological reasons but most people don't do things for ideology (note the emphasis on most). They do them for fun, profit, fame, convenience, or some other less noble goal.
As for the piece of hardware itself, hmmm, not sure why I would want to buy an overpriced and function limited and incompatible device.
Attending a conference (computer science or otherwise) doesn't mean much. You get to travel, stay at a fancy hotel (or a youth hostel if your university is poor) and present things. So what! There's that extra line on your CV.
It's worth it, though, if the people attending the conference are experts and you manage to discuss with them. Or if others see your work and build on it. Or if your work gets cited a lot as a result of attending the conference. Or if you manage to start an exciting joint research project. I've been to about 50 conferences so far and have published over 100 papers and the good ones are really worth it.
I'm not so sure about CCSC, though. Beyond that, I'm not a big fan of PhD conferences or sessions. If the work is good, everyone will want to hear it, so it'll be featured in the main conference anyway. If it isn't, having a special session won't help you.
Besides a) attenuation due to hand holding and b) change of the antenna characteristics due to bridging there's a third problem which really exacerbates the first two: the antenna of the iPhone 4G is highly directional. In other words, it matters a LOT which way you point the phone. Sometimes even small changes around it can make a big difference in terms of whether you get data or not.
You can test this out (assuming you've got access to an iPhone 4G) by running a speed test application (there are plenty in the App Store) while holding / pointing the phone in different ways. I can trigger signal loss even without holding the phone. No bumper whatsoever is going to fix that problem and this is plain and simple bad antenna design. I lose a lot more data when streaming radio on the 4G than what the 3G did even though the bandwidth is (potentially) much higher.
What the guys at NeuroSky are describing is complete vaporware. I work with brain signal data myself and know quite a few people who do. Basically, at present there are two methods that kind of work:
You implant a bunch of electrodes into a person's brain. See Michael Black's work (Brown University) on analyzing this data. You get roughly 30 bit per minute out of this. With some training a bit more. This is done for people who are seriously disabled, i.e. quadruplegics where you implant the electrodes in the motor cortex (useless for people who cannot move their limbs).
An alternative is to use EEGs. They usually come with about 100 electrodes, take an hour to put on and require lots of conductive gel. For instance Klaus Muller's group (Fraunhofer Institute Berlin) does such work. They get up to 20 bit per minute data rates. And yes, you can play simple games (they've got a cool demo of a person playing pong using the electrodes).
The big caveat is that there's just absolutely no way you can put a few electrodes onto your brain and get the information out that the NeuroSky people are claiming. The entire stuff looks really fishy, when you check their homepage http://www.neurosky.com/. Pretty much no information on who does the work, what their technology is, etc.
It's amazing how PWC and KPMG manage to create a market for themselves - the validation of spreadsheets. And it's amazing how this rather huge (and in my opinion concocted) figure is now quoted in many news outlets. There's no way the costs for spreadsheet errors are in this order of magnitude.
What we're seeing here is a really well executed PR maneuver by consulting companies.
As some users before, I've used a Commodore C128, then CP/M, DOS 3.xx after it became available, Geos, Windows 3.1 onwards, Mac OS 7, Linux since 0.99 using various window managers, such as twm, fvwm, KDE since 1.x up until 3.0 and lately OS X 10.1 until 10.3. And this mainly for programming, web development, scientific writing (LaTeX), web surfing, recently much of Office documents (unfortunately), and for entertainment.
As a desktop platform I must admit that OS X beats the rest hands down. And the reasons are stability and integration.
Not having to fiddle with XF86Config when plugging in a new external monitor helps.
Not having to recompile the kernel from source once you get an obscure USB or Firewire drive helps.
Not having to mess with the network configuration scripts as soon as you visit another lab helps.
Having a unified user interface helps (you save lots of time using the keyboard if you know that Command-Q quits every program, that Command-S saves files, that Command-N opens a new document, etc.).
Having Emacs bindings in forms like the one for posting on Slashdot helps.
Being able to watch a DVD without much fuss helps.
Being albe to sort my music and my photos easily helps.
Being able to run the system without crashes (BSD underbelly) helps.
Being able to use the shell and all UNIX tools helps.
Being able to open my laptop and to continue working within 2s helps. Especially if you spend lots of time in airports.
I'm not saying that none of this could be done under Linux. For almost every one of those items there's a tool that would allow me to do this. But this means that I have to go and configure it. It means that I have to spend my time on fixing it. For sure, Linux could do it. But it would mean that someone would have to produce a system that really works. Not just 90%, not just for most of the cases, not just something that nerds and geeks like me can use.
Switching from Linux (after 10 years of use) to OS X was a matter of 2 days of inconvenience. When my Mac broke and I had to switch back to my old system temporarily, it took me almost a week getting used to all the disincongruous interface tweaks again. And it's the first OS I'm not swearing at.
In particular, if you want a Unix capable laptop, you'd probably spend over 3 weeks tinkering with Linux until the system works properly (and it might not for recent hardware unless you hack it yourself - software modems, suspend to disk, wireless access, switching to external display, good power management). In a commercial environment that isn't worth it. Think your salary for three weeks vs. the price of the computer. And that's why in computer science you now see so many mac laptops when you go to conferences...
Sure, it works for a reasonably low bitrate. And it works for a low number of users. But it won't give you DSL-grade speed. Just a simple calculation: building the satellite and launching it will be at least $75,000. The receiver equipment, is at least $500 if you want tracking. A DSL line with 1MBit bandwidth, on the other hand costs $50/month and you want an amortization within 3 years. So we're talking about roughly 60 subscribers. This means that, with overheads (CDMA, TDMA, etc.), you'd need at least 100MBit bandwidth. FYI - terrestrial wireless lan gets you less than 10MBit.
And as for the off the shelf issue - do you remember the problems the radio satellite Oscar had?
The compression part does not impress me to the least. Sure, you can achieve fabulous rates on texts, and probably also on raw images, but most of the data is already compressed so you can't compress it any further.
Just think about mp3 files,.zipped files, jpeg, mpeg, pdf. Most content file formats are compressed, the big exception being executable files, text files, and almost all older Microsoft file formats (DOC, PPL, XLS, WMF, BMP). So compression won't help you.
Moreover, if you bild compression into the hardware, you have unpredictable storage information.
Even with the power of Open Source you still have to fight with physics. Here's a few reasons why it won't work.
Off the shelf hardware is not hardened against radiation, hence it'll just break, no matter what operating system you have on the satellite. Guess why ESA, NASA, and others are using special CPU, RAM and logic chips (tiny, slow, expensive) in their satellites.
Laser communication in space is a bad idea. also for big satellites - the positioning requirements are way to high. We're talking about inches rather than miles positioning precision for the beam. In other words, you won't be able to talk to your satellite. And if so, only very few people at the same time.
Linux or BSD may be good for earthbound systems but they're way too big for the small custom systems in satellites. Think of Linux on a Z80 or less.
You won't get enough output power from the satellite to detect it reliably on the earth - this means big and expensive receivers.
If you have a look at Intel's website, i.e. read before you post, you can see what kind of applications Intel has in mind. It's routers, controllers, etc. This kind of hardware hardly needs floating point performance at all, but lots of integer throughput. and for the few things that require a fast FPU you can always code it up in software. Just think of integer and floating point DSPs.
Besides, the Motorola Dragonball of the Palm isn't that fast in FP performance either. And you can always buy the coprocessor if need be. I think it's a great processor.
Patenting and licensing issues aside (not everything that is good is free and not everything that is free is good either) the claims of MP3pro sound very good. I think it will be successful, and here's why:
The fact that you can get sound quality that at 64kbit sounds like a 128kbit MP3 means that you get higher quality at 128kbit, too. In other words, your 128kbit MP3pro might sound like 200kbit MP3 or similar. So a more efficient codec is useful in any case, also since it most likely will give you CD quality at a lower bitrate.
VBR is usually independent of a better compression scheme. I guess, you can employ it for MP3pro, too, without too much trouble.
The format is backwards compatible to MP3, this means you can keep on using all your old MP3 hardware. This is the killer feature that no other format has. Just think of going from black and white TV to color TV, vs. color TV to HDTV (the latter requires that you buy a new TV set).
DjVu was not developed by the Olivetti and
Oracle labs in Cambridge. Instead the guys who did it were at Redbank, NJ.
Some of the members of the team include Leon Bottou, Yann LeCun (the guy who was one of the few inventors of Backprop), Yoshua Bengio, Patrick Haffner, Patrice Simard, and Larry Rabiner. I know those guys and they're very good. Don't know why ATT ultimately sold this product, though.
As far as the text compression goes - it works by clustering individual characters into subgroups and using the latter as a highlevel compression scheme plus encoding the differences. So you could even 'edit' the text (they did it but didn't release it for some reason). Anyway, it's pretty cool stuff.
Your point about minimum weights for each state are quite valid. However, there's an easy fix to solve, or at least alleviate the problem of a candidate winning the popular vote but failing to gain a sufficient number of candidates:
rather than assigning all the delegates to the candidate who gets the majority of the votes, split up the delegates proportionally to the fractions of the voters, probably with a slight bias towards the guy who got the majority (there are algorithms for this). It's a little bit like converting doubles into singles on a computer - you get roundoff errors. And by using a higher precision you can avoid that.
First off, there's always the update directory for
SuSE distributions which allows you to update your
packages automatically from yast. So it isn't quite that bad.
Secondly, The Kompany will be releasing a similar update tool as Helix/GNOME really soon for KDE.
Thirdly, if you use Sax2, it automatically does configure itself for the highest resolution that the monitor supports (i tried it with a TNT2 clone and it worked beautifully).
And of course, for the IP you can use DHCP. Works out of the box (at least for SuSE, but I don't know about Redhat).
If you've got a bunch of workstations to install, Alice is definitely the way to go. It took me a day to figure out how it works (the docu on alice is still quite bad) but once it's up and running you don't want to go back any more. Including all custiomization it takes less than 20 minutes to set up a box.
And SuSe 7.0 has been rock solid so far and quite bugfree (the only quirk I found so far was the fact that if you install PDFLaTeX later, it doesn't regenerate the hash table for fonts - which is easy to fix by running texhash). And configs are really easy - all the relevant info is in rc.config (and not all over the place as in rht).
Anyway, the professional version is worth the money - you're getting 4 handbooks with it. You'd probably spend the same money on linux book otherwise (without getting a new distro).
Your argument leads itself ad absurdum as well. The key question is who forfeited the right to use the code (which is not proven after all, but that's a different story and not relevant here).
If it is (a) the individual coder, then simply ask your buddy to 're-release' the code and you're fine since he is not affected by the GPL violation and nor is the 'pure' code either.
If it is (b) the KDE foundation (if such a thing actually exists) then still the coders could do so in their own right since they first and foremost own the copyrights to their code.
Option (c) is to create a legal entity on paper simply for doing this job. Subsequently the new code can be handed back to the original coders.
Any of these options are pretty much braindead and should not really be necessary. Some people think that the GPL is the One Free License. The QPL had nice bits, too, such as the request to release changes done in-house.
The key problem is most likely not only the mechanical one of soldering in the new chip but also that you have to be ultra careful not to zap the chips by electrostatic charges. So you need antistatic wrist straps and a place to attach them.
I seriously doubt they did it in the bar in Akihabara which they mention on their Japanese page. This area of Tokyo is one of the most lively places you can imagine (only Shibuya or one of the major subways stations are similar).
Besides that, you really need a very steady hand to solder pins that have a 1.27mm (i.e. 1/20") grid, hence roughly 0.5mm distance between the pins.
But with the right tools it's ok. I once had to build a pre-series demo model using SMD chips and it worked fine (but took forever).
Sure, there may be bad standards around, but at least you have something against which you can compare your code or your application.
For instance, you can use an old windows driver or old windows code (or even dos) under all different versions of Windoze. This may not always work perfectly and sometimes will hang the system but at least in 80% of all cases you can. Now try to do the same thing with a device driver you got for linux 2.2.12 and use it with 2.2.16 (I'm not even talking about major releases). In about 80% of all cases it will hang the system or not even load since lots of the interfaces changed.
You may suggest to recompile the drivers. Good idea, unless you've got a binary only driver from a company that got sued into oblivion as happened with Aureal (they won the lawsuit but went broke on expenses for their attorneys).
In other words, what open source needs now are open standards and with that I mean standards that are documented and that do not change whenever someone decides he wants to add on yet another feature. Or at least keep them backwards compatible.
This is going to get moderated down...
It seems that it has become extremely fashionable to criticize every single detail of whatever might not be absolutely politically correct about QT and KDE. On the other hand, whatever GNOME does is welcomed, even if it is, to say the least, as inflammatory as any of the KDE/Trolltech comments (e.g. Icaza's Unix sucks comments).
Trolltech is a software company that has to make money to survive. And they are writing fairly decent software for that, and provide it for free for non-commercial applications. All they get in return (from the slashdot crowd) is flames. Don't you think that this will make any other company think twice before they provide something for free for Open Source development?
As for the Gnome `dissing': Trolltech already took it off the website in no time. Please read before posting.
As for QT designer: sure there's more toolkits available (Kylix, KDevelop, etc.). But, hey, what's so bad about having yet another one? And for commercial development, 2.5k$ is nothing. A good coder costs heaps more than that (for instance MATLAB, a numerical simulations environment costs $10k for industry and $1k for universities).
So stop complaining and start coding if you want to make a point.
Don't get too excited about Linux supremacy. MS has still the lion's share of the market and most customers wouldn't buy a PC if they couldn't run M$ products on it.
Just have a look at what happened to Alpha. Since most programs didn't run natively on it (only via an x86 emulator) nobody bought the boxes. They're great for big numerical simulations, if you're willing to code everything on your own, though.
They look too blurred to me. Maybe they were doctored with Photoshop/Gimp. At least it isn't that difficult to take a good picture. In particular, since all four pics were so blurred that you couldn't really read anything (maybe except for a Tux lookalike).
I looked at the jpg file and the pictures seem to be taken with an Olympus SR85 (but that might be fake). These things should be able to give decent pictures even at short distances.
Besides that, why would you want to have Linux on a PDA in the first place? You don't need all the bells and whistles that come with it (the same is true for WinGE).
You can avoid this problem quite simply by encrypting the data or cryptographically signing the client or its output. Just remember the "blessed" binaries used in some first person shooter games on the web in order to avoid cheating by hacked clients.
A close look at the website of the NIC box reveals that
they are using a PCTEL softmodem. Anyone knows whether
such linux drivers are generally available (PCTEL said they
wouldn't sell to the public)?
Slashdot Mirror
I know this is going to be a flamebait. But before you flame me, consider the following: I'm researcher and get paid for what I do. I've released quite a few codes as open source and invented a bunch of algorithms which are not patented and used in many applications (think email spam filter, face recognition, etc.). And I've worked in industry and academia. For almost two decades. So I know both open and closed source.
First off, ideas have value. As in Dollar value. Take NVIDIA for instance - they don't have a semiconductor fab, so they send their chip layout to a place like TSMC or Global Foundry or Samsung or any other place to have their files turned into chips. These places are like modern printing presses. If their mask, vhdl or layout information were open source they wouldn't be able to reap the benefit from their investment into building the next generation of chips. Or as a more extreme case, take ARM. They design processor cores and license the microarchitecture to other (possibly fabless) design companies such as Apple which, in turn, tweak the design, add more stuff to it, and then ship it to the foundries. In other words, all the good stuff is in the plans, much less in the actual hardware.
So, designing an open source CPU is probably not going to work. Why not? Well, unlike with software, there's a massive barrier to entry. Talk Millions of Dollars rather than a few hundred to buy a laptop and install some version of GCC on it. Few users can afford this. This pretty much kills the model where many users take advantage of a good idea and share it to make it better. Yes, there are good ideological reasons but most people don't do things for ideology (note the emphasis on most). They do them for fun, profit, fame, convenience, or some other less noble goal.
As for the piece of hardware itself, hmmm, not sure why I would want to buy an overpriced and function limited and incompatible device.
Attending a conference (computer science or otherwise) doesn't mean much. You get to travel, stay at a fancy hotel (or a youth hostel if your university is poor) and present things. So what! There's that extra line on your CV.
It's worth it, though, if the people attending the conference are experts and you manage to discuss with them. Or if others see your work and build on it. Or if your work gets cited a lot as a result of attending the conference. Or if you manage to start an exciting joint research project. I've been to about 50 conferences so far and have published over 100 papers and the good ones are really worth it.
I'm not so sure about CCSC, though. Beyond that, I'm not a big fan of PhD conferences or sessions. If the work is good, everyone will want to hear it, so it'll be featured in the main conference anyway. If it isn't, having a special session won't help you.
Besides a) attenuation due to hand holding and b) change of the antenna characteristics due to bridging there's a third problem which really exacerbates the first two: the antenna of the iPhone 4G is highly directional. In other words, it matters a LOT which way you point the phone. Sometimes even small changes around it can make a big difference in terms of whether you get data or not.
You can test this out (assuming you've got access to an iPhone 4G) by running a speed test application (there are plenty in the App Store) while holding / pointing the phone in different ways. I can trigger signal loss even without holding the phone. No bumper whatsoever is going to fix that problem and this is plain and simple bad antenna design. I lose a lot more data when streaming radio on the 4G than what the 3G did even though the bandwidth is (potentially) much higher.
What the guys at NeuroSky are describing is complete vaporware. I work with brain signal data myself and know quite a few people who do. Basically, at present there are two methods that kind of work:
You implant a bunch of electrodes into a person's brain. See Michael Black's work (Brown University) on analyzing this data. You get roughly 30 bit per minute out of this. With some training a bit more. This is done for people who are seriously disabled, i.e. quadruplegics where you implant the electrodes in the motor cortex (useless for people who cannot move their limbs).
An alternative is to use EEGs. They usually come with about 100 electrodes, take an hour to put on and require lots of conductive gel. For instance Klaus Muller's group (Fraunhofer Institute Berlin) does such work. They get up to 20 bit per minute data rates. And yes, you can play simple games (they've got a cool demo of a person playing pong using the electrodes).
The big caveat is that there's just absolutely no way you can put a few electrodes onto your brain and get the information out that the NeuroSky people are claiming. The entire stuff looks really fishy, when you check their homepage http://www.neurosky.com/. Pretty much no information on who does the work, what their technology is, etc.
It's amazing how PWC and KPMG manage to create a market for themselves - the validation of spreadsheets. And it's amazing how this rather huge (and in my opinion concocted) figure is now quoted in many news outlets. There's no way the costs for spreadsheet errors are in this order of magnitude.
What we're seeing here is a really well executed PR maneuver by consulting companies.
As some users before, I've used a Commodore C128, then CP/M, DOS 3.xx after it became available, Geos, Windows 3.1 onwards, Mac OS 7, Linux since 0.99 using various window managers, such as twm, fvwm, KDE since 1.x up until 3.0 and lately OS X 10.1 until 10.3. And this mainly for programming, web development, scientific writing (LaTeX), web surfing, recently much of Office documents (unfortunately), and for entertainment.
As a desktop platform I must admit that OS X beats the rest hands down. And the reasons are stability and integration.
- Not having to fiddle with XF86Config when plugging in a new external monitor helps.
- Not having to recompile the kernel from source once you get an obscure USB or Firewire drive helps.
- Not having to mess with the network configuration scripts as soon as you visit another lab helps.
- Having a unified user interface helps (you save lots of time using the keyboard if you know that Command-Q quits every program, that Command-S saves files, that Command-N opens a new document, etc.).
- Having Emacs bindings in forms like the one for posting on Slashdot helps.
- Being able to watch a DVD without much fuss helps.
- Being albe to sort my music and my photos easily helps.
- Being able to run the system without crashes (BSD underbelly) helps.
- Being able to use the shell and all UNIX tools helps.
- Being able to open my laptop and to continue working within 2s helps. Especially if you spend lots of time in airports.
I'm not saying that none of this could be done under Linux. For almost every one of those items there's a tool that would allow me to do this. But this means that I have to go and configure it. It means that I have to spend my time on fixing it. For sure, Linux could do it. But it would mean that someone would have to produce a system that really works. Not just 90%, not just for most of the cases, not just something that nerds and geeks like me can use.Switching from Linux (after 10 years of use) to OS X was a matter of 2 days of inconvenience. When my Mac broke and I had to switch back to my old system temporarily, it took me almost a week getting used to all the disincongruous interface tweaks again. And it's the first OS I'm not swearing at.
In particular, if you want a Unix capable laptop, you'd probably spend over 3 weeks tinkering with Linux until the system works properly (and it might not for recent hardware unless you hack it yourself - software modems, suspend to disk, wireless access, switching to external display, good power management). In a commercial environment that isn't worth it. Think your salary for three weeks vs. the price of the computer. And that's why in computer science you now see so many mac laptops when you go to conferences ...
And as for the off the shelf issue - do you remember the problems the radio satellite Oscar had?
Just think about mp3 files, .zipped files, jpeg, mpeg, pdf. Most content file formats are compressed, the big exception being executable files, text files, and almost all older Microsoft file formats (DOC, PPL, XLS, WMF, BMP). So compression won't help you.
Moreover, if you bild compression into the hardware, you have unpredictable storage information.
Off the shelf hardware is not hardened against radiation, hence it'll just break, no matter what operating system you have on the satellite. Guess why ESA, NASA, and others are using special CPU, RAM and logic chips (tiny, slow, expensive) in their satellites.
Laser communication in space is a bad idea. also for big satellites - the positioning requirements are way to high. We're talking about inches rather than miles positioning precision for the beam. In other words, you won't be able to talk to your satellite. And if so, only very few people at the same time.
Linux or BSD may be good for earthbound systems but they're way too big for the small custom systems in satellites. Think of Linux on a Z80 or less.
You won't get enough output power from the satellite to detect it reliably on the earth - this means big and expensive receivers.
Besides, the Motorola Dragonball of the Palm isn't that fast in FP performance either. And you can always buy the coprocessor if need be. I think it's a great processor.
Just my 5c.
The fact that you can get sound quality that at 64kbit sounds like a 128kbit MP3 means that you get higher quality at 128kbit, too. In other words, your 128kbit MP3pro might sound like 200kbit MP3 or similar. So a more efficient codec is useful in any case, also since it most likely will give you CD quality at a lower bitrate.
VBR is usually independent of a better compression scheme. I guess, you can employ it for MP3pro, too, without too much trouble.
The format is backwards compatible to MP3, this means you can keep on using all your old MP3 hardware. This is the killer feature that no other format has. Just think of going from black and white TV to color TV, vs. color TV to HDTV (the latter requires that you buy a new TV set).
3bits means 8 different states. This is quite different from 3 states, which would mean log(3)/log(2) = 1.5850 bits per pixel.
Some of the members of the team include Leon Bottou, Yann LeCun (the guy who was one of the few inventors of Backprop), Yoshua Bengio, Patrick Haffner, Patrice Simard, and Larry Rabiner. I know those guys and they're very good. Don't know why ATT ultimately sold this product, though.
As far as the text compression goes - it works by clustering individual characters into subgroups and using the latter as a highlevel compression scheme plus encoding the differences. So you could even 'edit' the text (they did it but didn't release it for some reason). Anyway, it's pretty cool stuff.
rather than assigning all the delegates to the candidate who gets the majority of the votes, split up the delegates proportionally to the fractions of the voters, probably with a slight bias towards the guy who got the majority (there are algorithms for this). It's a little bit like converting doubles into singles on a computer - you get roundoff errors. And by using a higher precision you can avoid that.
Secondly, The Kompany will be releasing a similar update tool as Helix/GNOME really soon for KDE.
Thirdly, if you use Sax2, it automatically does configure itself for the highest resolution that the monitor supports (i tried it with a TNT2 clone and it worked beautifully).
And of course, for the IP you can use DHCP. Works out of the box (at least for SuSE, but I don't know about Redhat).
And SuSe 7.0 has been rock solid so far and quite bugfree (the only quirk I found so far was the fact that if you install PDFLaTeX later, it doesn't regenerate the hash table for fonts - which is easy to fix by running texhash). And configs are really easy - all the relevant info is in rc.config (and not all over the place as in rht).
Anyway, the professional version is worth the money - you're getting 4 handbooks with it. You'd probably spend the same money on linux book otherwise (without getting a new distro).
If it is (a) the individual coder, then simply ask your buddy to 're-release' the code and you're fine since he is not affected by the GPL violation and nor is the 'pure' code either.
If it is (b) the KDE foundation (if such a thing actually exists) then still the coders could do so in their own right since they first and foremost own the copyrights to their code.
Option (c) is to create a legal entity on paper simply for doing this job. Subsequently the new code can be handed back to the original coders.
Any of these options are pretty much braindead and should not really be necessary. Some people think that the GPL is the One Free License. The QPL had nice bits, too, such as the request to release changes done in-house.
Ok, sorry, you're probably right. Reading Japanese really helps ;).
I seriously doubt they did it in the bar in Akihabara which they mention on their Japanese page. This area of Tokyo is one of the most lively places you can imagine (only Shibuya or one of the major subways stations are similar).
Besides that, you really need a very steady hand to solder pins that have a 1.27mm (i.e. 1/20") grid, hence roughly 0.5mm distance between the pins.
But with the right tools it's ok. I once had to build a pre-series demo model using SMD chips and it worked fine (but took forever).
For instance, you can use an old windows driver or old windows code (or even dos) under all different versions of Windoze. This may not always work perfectly and sometimes will hang the system but at least in 80% of all cases you can. Now try to do the same thing with a device driver you got for linux 2.2.12 and use it with 2.2.16 (I'm not even talking about major releases). In about 80% of all cases it will hang the system or not even load since lots of the interfaces changed.
You may suggest to recompile the drivers. Good idea, unless you've got a binary only driver from a company that got sued into oblivion as happened with Aureal (they won the lawsuit but went broke on expenses for their attorneys).
In other words, what open source needs now are open standards and with that I mean standards that are documented and that do not change whenever someone decides he wants to add on yet another feature. Or at least keep them backwards compatible.
It seems that it has become extremely fashionable to criticize every single detail of whatever might not be absolutely politically correct about QT and KDE. On the other hand, whatever GNOME does is welcomed, even if it is, to say the least, as inflammatory as any of the KDE/Trolltech comments (e.g. Icaza's Unix sucks comments).
Trolltech is a software company that has to make money to survive. And they are writing fairly decent software for that, and provide it for free for non-commercial applications. All they get in return (from the slashdot crowd) is flames. Don't you think that this will make any other company think twice before they provide something for free for Open Source development?
As for the Gnome `dissing': Trolltech already took it off the website in no time. Please read before posting.
As for QT designer: sure there's more toolkits available (Kylix, KDevelop, etc.). But, hey, what's so bad about having yet another one? And for commercial development, 2.5k$ is nothing. A good coder costs heaps more than that (for instance MATLAB, a numerical simulations environment costs $10k for industry and $1k for universities).
So stop complaining and start coding if you want to make a point.
Don't get too excited about Linux supremacy. MS has still the lion's share of the market and most customers wouldn't buy a PC if they couldn't run M$ products on it.
Just have a look at what happened to Alpha. Since most programs didn't run natively on it (only via an x86 emulator) nobody bought the boxes. They're great for big numerical simulations, if you're willing to code everything on your own, though.
Are the screenshots for real?
They look too blurred to me. Maybe they were doctored with Photoshop/Gimp. At least it isn't that difficult to take a good picture. In particular, since all four pics were so blurred that you couldn't really read anything (maybe except for a Tux lookalike).
I looked at the jpg file and the pictures seem to be taken with an Olympus SR85 (but that might be fake). These things should be able to give decent pictures even at short distances.
Besides that, why would you want to have Linux on a PDA in the first place? You don't need all the bells and whistles that come with it (the same is true for WinGE).
You can avoid this problem quite simply by encrypting the data or cryptographically signing the client or its output. Just remember the "blessed" binaries used in some first person shooter games on the web in order to avoid cheating by hacked clients.
A close look at the website of the NIC box reveals that they are using a PCTEL softmodem. Anyone knows whether such linux drivers are generally available (PCTEL said they wouldn't sell to the public)?