Anyone who believes that machines can replace learning a language has clearly never left his country or spent more than a week abroad. There are technical and cultural issues that render such statements nonsense.
Technical: - you need to speak like another machine for these systems to recognize what you say. Start putting some accent (like the different Latin-Spanish versions), or dialects (like in Germany or China), slang, and the model breaks quickly. - no system is able to mix languages. And you need this. It is common to mix languages with certain words, street names, person names, etc. from other languages. - street language. Even if the sound recognition were perfect, no machine translator can possibly translate what you hear on the street.
Cultural: - go to a sales meeting and you are trying to sell your business services to a customer using your voice as translation. Your competitor speaks the language fluently, using idioms and other tricks. Guess who gets the deal. - pick-up a girl in Italy using a phone/voice translation and I will aplaud you. - attend university abroad using your tech-device. - tell a Joke to your phone, hoping that its translation will make your foreign friends laugh.
the list is endless. So is this a good invention? Yes. Will it work? Maybe in the future for some limited purposes. Will it replace learning languages? Heck no.
just say that we use version control to do what we want:
- using Git/Mercurial is like walking naked on the beach - using svn is walking naked at home - using Clear Case is going through airport scanning and being detained 30 min for having a nail cutter, followed by anal examination - using Visual Source Safe is terrorism. It's lie having your balls hit repeatedly, by something like the door of a submarine.
In Europe it is common for people to get contracts to just subsidise the phone, but not the data and calls. You often hear people saying "I've bought my high-end phone for 50 Euros", but then pay 20 Euros per month on 24 hour contract, and 40 cents/minute and 20c/SMS. So it is not like in the US that you only get a full package. You can also get full packages in Europe, but once again this is not the rule.
A more economical approach at the moment, at least in Germany, is to buy your own phone at full price and use pre-paid. It is funny, but pre-paid is way cheaper than the contracts and you are not tight to a carrier for two years. This is for me crucial because ALL carriers claim that you can't use your phone for VoIP, and sometimes even for IM. None of them block them that I know (I think the European Laws would hunt them), but I don't wanna have my Whatsapp block one day and be stucked with 12 more months to go.
You can also get pretty good deals for full packages through your employer. If you can live with 24 month contracts, these are sometimes better then pre-paid+full priced phone deals.
you also need to have NFC enabled on your Galaxy for this to work.
No, you don't. If you take a minute to RTFA you'll see this:
The attack isn’t limited to NFC though; it can also be abused via other attack vectors, such as malicious websites or email attachments.
Yes, you do. What you are describing is a different way to accomplish the attack. As an end user, I don't care if the underlying exploit is similar, I only care about how I can be affected by it. This leads to the next point.
They chose to use NFC for the novelty effect.
No, they've chosen NFC because now more phones have it, but mostly because it allows accomplishing the attack without any user intervention at all. People could avoid getting hacked from visiting malicious websites, simply by limiting themselves to trusted sites. Most people only frequent their usual places. But the NFC is a hidden vector that many users are not even aware of.
As I've mentioned in my first post, I could live with an NFC or browser vulnerability, but not with a tethering one. Other people will think the opposite. At the end of the day, these news make wish you didn't depend on your cell phone so much, because there are always security holes in there.
I find it funny when automotive industry push to connect their cars to the network, as if they could do any better.
you also need to have NFC enabled on your Galaxy for this to work. NFC is enabled by default, sure. But it can be disabled easily. I also find myself living happily without NFC, but not without tethering, which I use daily during my bus commute.
So my point is that both vulnerabilities suck, and which one sucks the most depends solely on your use-case. There is no point in saying that one device is more secure than the other, both Apple and Google seem to suck big time here. You should not store any sensitive data on your phone.
As I've posted above, you have a valid point and I won't argue whether this was fair or not. What I was trying to convey is that if teachers are gonna grade like computers without a human factor in there, then I might as well just take online classes for a fraction of the price.
All valid points. And I'm not arguing either that I didn't deserve that D, I'm trying to not judge myself here. What I'm arguing is that if the human/fuzzy/sensitive factor can't be used for teaching, then you can replace teachers with computers. So my point here is: either be different from a computer, or let me take online classes from home for a fraction of the price.
everyone in Argentina is proud about the Hand of God. It showed two things: - If he wanted, Maradona was able to fool the referee and make fun of the brits, despite them having stolen their island shortly before. - If he wanted, Maradona could just a few minutes later make one of the best goals in history. Of course the brits don't acknowledge it and just whine about the first one.
I don't wanna mix technology, sports, and politics here. I just wanted to point out that no one in Argentina is ashamed of the hand of god.
Mod parent up, he's key on. I remember an issue with a professor in one of my C++ classes, which happened to include a large programming project. The project took about four weeks of intensive programming, and I was really proud of the quality of my code, comments, structure, etc. Only problem was that in one section we had to determine the actual type of an object using dynamic_cast after having received a base type object. We had like 10 derived objects and I've used copy paste to make life easier, but forgot to modify one entry with the appropriate type. That is, ONE word was wrong. My mistake failed in one of their tests (which I didn't have in advance), which cascaded four output missmatches. This ONE word cost me 40 points out of 100, ending up with a D for this project. One word, lots of effort. I've talked to the professor and his answer was a lame "if I fix your grade, I need to fix everyone's".
When I was a TA during grad school, I always looked at the work flow. If a student made a mistake in part (a) of a problem, I didn't simply give him zero points for parts (b) (c) and (d) that used it as a base. Instead, I've assumed that part (a) was right and looked at the process. It took me more time to grade, sure. But it is fair and if a teacher can't contribute with some human touch, let's just replace them with computers.
Natvteq has actually very good maps, particularly in Europe. The reason why your map looks bad is most likely due to your particular gear vendor. Garmin, Becker, Blaupunkt, Falk, etc. they buy maps from Navteq or Teleatlas, and they compile it for whatever their main goal and budget. In order to reduce the map size and save money, they compress the data using a battery of techniques. One common technique is decimation, where they simply remove geometry points to save space, leaving mostly the ones that represent intersections and a few in the middle. POI suffer as well.
So please don't be too fast in blaming a map vendor, where the fault is almost certainly from your navigation system vendor.
awesome and thanks for the clarification. Now what puzzles me here: if you really care about people using it: why is the only supported Ubuntu Version 10.04? Wouldn't it be meaninfull to support a later release?
I'm glad that Microsoft is working on Linux, but I was just wondering whether I want to update. I use Skype on both Linux and Windows and to be honest, I prefer the old version from Linux. Sure: it crashes more often, and has somehow less features. But for what I mostly do, it works ok. And the interface is much less cluttered, I can quit the app without having to read a manual, no ads, really slick.
I haven't looked at the new Skyp 4 yet, but I don't find it unlikely that I switch back to the old version at some point. One nice thing about Skype being so closed and proprietary is that the chances of seeing a Linux exploit based on Skype are extremely low.
Ubuntu gives you Skype through Canonical partners repository. I think you are asked whether you want it there during the installation, in the same window where they ask about multimedia support and updates during installation. Downside is that we will probably not see the new version for while.
Absolutely correct. OSM accuracy is very good, indeed. The only "problem" with it is that it's a best effort, and it doesn't have a standard way of doing it. For example, when you trace a 4-lane highway, on which lane do you collect the data? If you cut a curve when driving on the mountains, how do you deal with it later? How do you decide where to place your shapepoints so that the error in the true curvature is below the required threshold? There are guidelines, and people should follow them. But if they don't so be it. Commercial map vendors do this for living and must follow their data-collection procedures, which are very well documented.
For many applications, OSM is awesome. But there is a group of applications, and I can only speak about the automotive sector, where paying for commercial maps is still worth it. Of course I don't mean navigation, where OSM is already good enough.
this is because the map you have on your navigation system has been optimized for navigation and highly compressed. You don't need this high accuracy for navigation. But if you bought the maps from Navteq or Teleatlas directly, you'd see that the accuracy is much better. Remember: maps is *not* just navigation. There are literally hundreds if not thousands of other applications that can benefit from digital maps.
This is a nice guess and usually works, but it is not guaranteed to give you 1m relative quality. You can have systematic errors that apply to all users that traveled the road that day, such as bad weather or ionosphere effects. But another problem is that users drive by car using particular patterns that do not necessarily match the true road geometry. Some examples of these are curve-cutting when driving, lane changes on highways, etc.
The approach you are describing would require more than 25 people and as a matter of fact, this is what Teleatlas (Tom Tom) does with their shared data. Nevertheless, it is a statistical best guess, and the method shall not be compared with a trace collected via DGPS. They are fundamentally different.
I have been working with digital map data for on advanced driver assistance applications for a few years, and my experience is the following:
Some applications want lots of data. They don’t care if it is perfect or not, such as whether there is a zebra crossing, a traffic light, a stair, a path for mountain bike but not for road bike, etc. One example of this is navigation: it doesn’t matter if the turn has an error of 10 meter, if it is 10 or 25 degrees to the right, etc.
Other applications they are fine with less data, but this must be absolutely accurate to within a meter. Examples of this are active-safety applications, such as map-based adaptive front lighting, curve warning, etc.
Some other applications are in the middle. They are not very sensitive, but annoying if incorrect: example of this is speed limit warning.
The biggest map vendors collect hundreds of attributes at very high quality. This is true particularly for low-number functional classes (highways and motorways). They often meet the 5-m absolute and 1-m relative accuracy for geometry.
It is very difficult for OSM to meet this high quality, specially because you need a differential GPS (DGPS) to collect these. That said, map vendors invest most of their effort on large important roads, while rural or off-roads have from low to very poor quality.
Moreover, one thing is the quality at which data is collected, and another one is the map quality. Vendors tend to decimate (strip-out) geometry points on non important roads in order to reduce the size of the map.
So to sum up: if you are on a motorway or highway, OSM probably won’t match the quality of Navteq, Teleatlas or Google. If you are on a rural area, off-road, bike trail, etc., OSM will probably kick everyone’s butt. Plus it is usually more up-to-date.
TomTom tries to close this gap with their community content, which I find very dishonest from them. They save millions by using people’s data, but they don’t pass these savings back to the consumer.
The article implies that if you go pre-paid, you have to put-up with a low-end phone. Why? Can't you get a top-end unlocked Samsung Galaxy S2 to use with your pre-paid? Sure, it will cost you 600 dollar. But you will not have to pay 80 dollar/month on your plan. And the best: if they screw you, you can just dump them and switch provider while keeping your phone.
The ability to change carriers easily is great for competition. Look at how it worked in Europe: you can get a line at zero cost per month. Add 4 euros per month and you have unlimited calls within your network. 10 euros/month for data plan. And some give you at the same time an airbag for a maximum of 40 euros/month, so you never pay more than that no matter what. If people moved out of the big carriers, same thing will happen in the US.
It might seam that it does what you say, but it doesn't. Please read the instructions on how the package upgrade process works. It essentially means creating a list of the applications you have installed, doing a reinstallation of mint, and then using apt-get to reinstall all those apps again. Sure: if you have your home on a different partition and you are very careful you could tell the installer to not format your home, and then you could kind of do an upgrade. But I'm not sure how seamless this would work with encrypted home, it will definitely not work with full disc encryption, RAID, LVM, etc, and you could easily mess up your whole installation and lose your data.
Compare this to the easy upgrade offered by Ubuntu since ages. I don't mean to rant: the points made by Clement in that page are valid: the upgrade process is not perfectly clean. But it does work most of the times (I have never had an issue with it), and it is definitely straightforward and the only feasible alternative to non Linux techies.
And in 6 months your top geeks will bitch about why they can't just upgrade like with other distros. In any case, make sure your machines are renewed every three years, or settle for the LTS.
Mint can be upgraded between releases. This makes them drastically different, up to the point that Mint is not an option for me. Their target users are as a result different.
And I don't mean to troll with that. I really like Mint and respect Clement's work. But I can't recommend Mint to everyone as much as Ubuntu.
I hope you are serious, because I really need Kubuntu. I'm not kidding, it is the only distro that satisfies my needs. I'll explain because I'm sure other people are on the same boat:
- Want KDE, of course - Easy to install. Must work out of the box - Installer must work. When I gave OpenSuse a shot, the installer messed up my truecrypt MBR that I use for Windows 7, although I told it to install GRUB on the/root partition. So much for OpenSuse - Up-to-date Software. I also like the ability to give new KDE releases a try without waiting for 3 months for them. KDE PPA is awesome - Easy upgrade every six months. An upgrade must work without backup/restore. I respect Clement's work on Mint very much, but for me, I don't want to backup and restore my system every six months. - Repository for Skype, proprietary drivers, etc.
Have a laptop, and it is sometimes used by different people at my home. This means: - Need encryption in case it gets stolen when I travel with it. Don't want full disc encryption, because that means typing two passwords. I want one single password. - Don't want container-based encryption, because that means preallocating space for each user. I want something that grows dynamically, based on whatever a user needs. Ubuntu's approach for home encryption using individual files is just what I want. I know is not good for sparse files, but that's not my use-case. - I have nvidia-optimus. So I need an easy way to install Bumbebee. Kubuntu gives you that. And nope, no BIOS switch so I need it in software
As far as I could tell, Kubuntu is the only distro that does what I need. So please don't let it die. Thanks
You think so? I just see vibrant colors, not 41megapixels detail. With a clear blue sky like that, and plenty of light, you minimize the problems of small sensors/lenses.
It is known that you can print up to A3 size with a 6 MP camera. So the only advantage would be when using digital zoom. If I zoom in those images I see a noisy sky. Sure, it can be a problem with jpg format. But in any case, the same picture could have been obtained with a 6MP camera.
Slashdot Mirror
Anyone who believes that machines can replace learning a language has clearly never left his country or spent more than a week abroad. There are technical and cultural issues that render such statements nonsense.
Technical:
- you need to speak like another machine for these systems to recognize what you say. Start putting some accent (like the different Latin-Spanish versions), or dialects (like in Germany or China), slang, and the model breaks quickly.
- no system is able to mix languages. And you need this. It is common to mix languages with certain words, street names, person names, etc. from other languages.
- street language. Even if the sound recognition were perfect, no machine translator can possibly translate what you hear on the street.
Cultural:
- go to a sales meeting and you are trying to sell your business services to a customer using your voice as translation. Your competitor speaks the language fluently, using idioms and other tricks. Guess who gets the deal.
- pick-up a girl in Italy using a phone/voice translation and I will aplaud you.
- attend university abroad using your tech-device.
- tell a Joke to your phone, hoping that its translation will make your foreign friends laugh.
the list is endless. So is this a good invention? Yes. Will it work? Maybe in the future for some limited purposes. Will it replace learning languages? Heck no.
just say that we use version control to do what we want:
- using Git/Mercurial is like walking naked on the beach
- using svn is walking naked at home
- using Clear Case is going through airport scanning and being detained 30 min for having a nail cutter, followed by anal examination
- using Visual Source Safe is terrorism. It's lie having your balls hit repeatedly, by something like the door of a submarine.
In Europe it is common for people to get contracts to just subsidise the phone, but not the data and calls. You often hear people saying "I've bought my high-end phone for 50 Euros", but then pay 20 Euros per month on 24 hour contract, and 40 cents/minute and 20c/SMS. So it is not like in the US that you only get a full package. You can also get full packages in Europe, but once again this is not the rule.
A more economical approach at the moment, at least in Germany, is to buy your own phone at full price and use pre-paid. It is funny, but pre-paid is way cheaper than the contracts and you are not tight to a carrier for two years. This is for me crucial because ALL carriers claim that you can't use your phone for VoIP, and sometimes even for IM. None of them block them that I know (I think the European Laws would hunt them), but I don't wanna have my Whatsapp block one day and be stucked with 12 more months to go.
You can also get pretty good deals for full packages through your employer. If you can live with 24 month contracts, these are sometimes better then pre-paid+full priced phone deals.
you also need to have NFC enabled on your Galaxy for this to work.
No, you don't. If you take a minute to RTFA you'll see this:
The attack isn’t limited to NFC though; it can also be abused via other attack vectors, such as malicious websites or email attachments.
Yes, you do. What you are describing is a different way to accomplish the attack. As an end user, I don't care if the underlying exploit is similar, I only care about how I can be affected by it. This leads to the next point.
They chose to use NFC for the novelty effect.
No, they've chosen NFC because now more phones have it, but mostly because it allows accomplishing the attack without any user intervention at all. People could avoid getting hacked from visiting malicious websites, simply by limiting themselves to trusted sites. Most people only frequent their usual places. But the NFC is a hidden vector that many users are not even aware of.
As I've mentioned in my first post, I could live with an NFC or browser vulnerability, but not with a tethering one. Other people will think the opposite. At the end of the day, these news make wish you didn't depend on your cell phone so much, because there are always security holes in there.
I find it funny when automotive industry push to connect their cars to the network, as if they could do any better.
you also need to have NFC enabled on your Galaxy for this to work. NFC is enabled by default, sure. But it can be disabled easily. I also find myself living happily without NFC, but not without tethering, which I use daily during my bus commute.
So my point is that both vulnerabilities suck, and which one sucks the most depends solely on your use-case. There is no point in saying that one device is more secure than the other, both Apple and Google seem to suck big time here. You should not store any sensitive data on your phone.
As I've posted above, you have a valid point and I won't argue whether this was fair or not. What I was trying to convey is that if teachers are gonna grade like computers without a human factor in there, then I might as well just take online classes for a fraction of the price.
All valid points. And I'm not arguing either that I didn't deserve that D, I'm trying to not judge myself here. What I'm arguing is that if the human/fuzzy/sensitive factor can't be used for teaching, then you can replace teachers with computers. So my point here is: either be different from a computer, or let me take online classes from home for a fraction of the price.
everyone in Argentina is proud about the Hand of God. It showed two things:
- If he wanted, Maradona was able to fool the referee and make fun of the brits, despite them having stolen their island shortly before.
- If he wanted, Maradona could just a few minutes later make one of the best goals in history. Of course the brits don't acknowledge it and just whine about the first one.
I don't wanna mix technology, sports, and politics here. I just wanted to point out that no one in Argentina is ashamed of the hand of god.
Mod parent up, he's key on. I remember an issue with a professor in one of my C++ classes, which happened to include a large programming project. The project took about four weeks of intensive programming, and I was really proud of the quality of my code, comments, structure, etc. Only problem was that in one section we had to determine the actual type of an object using dynamic_cast after having received a base type object. We had like 10 derived objects and I've used copy paste to make life easier, but forgot to modify one entry with the appropriate type. That is, ONE word was wrong. My mistake failed in one of their tests (which I didn't have in advance), which cascaded four output missmatches. This ONE word cost me 40 points out of 100, ending up with a D for this project. One word, lots of effort. I've talked to the professor and his answer was a lame "if I fix your grade, I need to fix everyone's".
When I was a TA during grad school, I always looked at the work flow. If a student made a mistake in part (a) of a problem, I didn't simply give him zero points for parts (b) (c) and (d) that used it as a base. Instead, I've assumed that part (a) was right and looked at the process. It took me more time to grade, sure. But it is fair and if a teacher can't contribute with some human touch, let's just replace them with computers.
Natvteq has actually very good maps, particularly in Europe. The reason why your map looks bad is most likely due to your particular gear vendor. Garmin, Becker, Blaupunkt, Falk, etc. they buy maps from Navteq or Teleatlas, and they compile it for whatever their main goal and budget. In order to reduce the map size and save money, they compress the data using a battery of techniques. One common technique is decimation, where they simply remove geometry points to save space, leaving mostly the ones that represent intersections and a few in the middle. POI suffer as well.
So please don't be too fast in blaming a map vendor, where the fault is almost certainly from your navigation system vendor.
awesome and thanks for the clarification. Now what puzzles me here: if you really care about people using it: why is the only supported Ubuntu Version 10.04? Wouldn't it be meaninfull to support a later release?
I'm glad that Microsoft is working on Linux, but I was just wondering whether I want to update. I use Skype on both Linux and Windows and to be honest, I prefer the old version from Linux. Sure: it crashes more often, and has somehow less features. But for what I mostly do, it works ok. And the interface is much less cluttered, I can quit the app without having to read a manual, no ads, really slick.
I haven't looked at the new Skyp 4 yet, but I don't find it unlikely that I switch back to the old version at some point. One nice thing about Skype being so closed and proprietary is that the chances of seeing a Linux exploit based on Skype are extremely low.
Ubuntu gives you Skype through Canonical partners repository. I think you are asked whether you want it there during the installation, in the same window where they ask about multimedia support and updates during installation. Downside is that we will probably not see the new version for while.
Absolutely correct. OSM accuracy is very good, indeed. The only "problem" with it is that it's a best effort, and it doesn't have a standard way of doing it. For example, when you trace a 4-lane highway, on which lane do you collect the data? If you cut a curve when driving on the mountains, how do you deal with it later? How do you decide where to place your shapepoints so that the error in the true curvature is below the required threshold? There are guidelines, and people should follow them. But if they don't so be it. Commercial map vendors do this for living and must follow their data-collection procedures, which are very well documented.
For many applications, OSM is awesome. But there is a group of applications, and I can only speak about the automotive sector, where paying for commercial maps is still worth it. Of course I don't mean navigation, where OSM is already good enough.
this is because the map you have on your navigation system has been optimized for navigation and highly compressed. You don't need this high accuracy for navigation. But if you bought the maps from Navteq or Teleatlas directly, you'd see that the accuracy is much better. Remember: maps is *not* just navigation. There are literally hundreds if not thousands of other applications that can benefit from digital maps.
This is a nice guess and usually works, but it is not guaranteed to give you 1m relative quality. You can have systematic errors that apply to all users that traveled the road that day, such as bad weather or ionosphere effects. But another problem is that users drive by car using particular patterns that do not necessarily match the true road geometry. Some examples of these are curve-cutting when driving, lane changes on highways, etc.
The approach you are describing would require more than 25 people and as a matter of fact, this is what Teleatlas (Tom Tom) does with their shared data. Nevertheless, it is a statistical best guess, and the method shall not be compared with a trace collected via DGPS. They are fundamentally different.
I have been working with digital map data for on advanced driver assistance applications for a few years, and my experience is the following:
Some applications want lots of data. They don’t care if it is perfect or not, such as whether there is a zebra crossing, a traffic light, a stair, a path for mountain bike but not for road bike, etc. One example of this is navigation: it doesn’t matter if the turn has an error of 10 meter, if it is 10 or 25 degrees to the right, etc.
Other applications they are fine with less data, but this must be absolutely accurate to within a meter. Examples of this are active-safety applications, such as map-based adaptive front lighting, curve warning, etc.
Some other applications are in the middle. They are not very sensitive, but annoying if incorrect: example of this is speed limit warning.
The biggest map vendors collect hundreds of attributes at very high quality. This is true particularly for low-number functional classes (highways and motorways). They often meet the 5-m absolute and 1-m relative accuracy for geometry.
It is very difficult for OSM to meet this high quality, specially because you need a differential GPS (DGPS) to collect these. That said, map vendors invest most of their effort on large important roads, while rural or off-roads have from low to very poor quality.
Moreover, one thing is the quality at which data is collected, and another one is the map quality. Vendors tend to decimate (strip-out) geometry points on non important roads in order to reduce the size of the map.
So to sum up: if you are on a motorway or highway, OSM probably won’t match the quality of Navteq, Teleatlas or Google. If you are on a rural area, off-road, bike trail, etc., OSM will probably kick everyone’s butt. Plus it is usually more up-to-date.
TomTom tries to close this gap with their community content, which I find very dishonest from them. They save millions by using people’s data, but they don’t pass these savings back to the consumer.
The article implies that if you go pre-paid, you have to put-up with a low-end phone. Why? Can't you get a top-end unlocked Samsung Galaxy S2 to use with your pre-paid? Sure, it will cost you 600 dollar. But you will not have to pay 80 dollar/month on your plan. And the best: if they screw you, you can just dump them and switch provider while keeping your phone.
The ability to change carriers easily is great for competition. Look at how it worked in Europe: you can get a line at zero cost per month. Add 4 euros per month and you have unlimited calls within your network. 10 euros/month for data plan. And some give you at the same time an airbag for a maximum of 40 euros/month, so you never pay more than that no matter what. If people moved out of the big carriers, same thing will happen in the US.
It might seam that it does what you say, but it doesn't. Please read the instructions on how the package upgrade process works. It essentially means creating a list of the applications you have installed, doing a reinstallation of mint, and then using apt-get to reinstall all those apps again. Sure: if you have your home on a different partition and you are very careful you could tell the installer to not format your home, and then you could kind of do an upgrade. But I'm not sure how seamless this would work with encrypted home, it will definitely not work with full disc encryption, RAID, LVM, etc, and you could easily mess up your whole installation and lose your data.
Compare this to the easy upgrade offered by Ubuntu since ages. I don't mean to rant: the points made by Clement in that page are valid: the upgrade process is not perfectly clean. But it does work most of the times (I have never had an issue with it), and it is definitely straightforward and the only feasible alternative to non Linux techies.
because with mint you can't really upgrade. Ugrading on mint just implies backup-reinstall-restore.
checkt it here
http://community.linuxmint.com/tutorial/view/2
And in 6 months your top geeks will bitch about why they can't just upgrade like with other distros. In any case, make sure your machines are renewed every three years, or settle for the LTS.
thanks but Fedora doesn't fullfill my requirements. Please take a look at my list above again.
Mint can be upgraded between releases. This makes them drastically different, up to the point that Mint is not an option for me. Their target users are as a result different.
And I don't mean to troll with that. I really like Mint and respect Clement's work. But I can't recommend Mint to everyone as much as Ubuntu.
I hope you are serious, because I really need Kubuntu. I'm not kidding, it is the only distro that satisfies my needs. I'll explain because I'm sure other people are on the same boat:
- Want KDE, of course /root partition. So much for OpenSuse
- Easy to install. Must work out of the box
- Installer must work. When I gave OpenSuse a shot, the installer messed up my truecrypt MBR that I use for Windows 7, although I told it to install GRUB on the
- Up-to-date Software. I also like the ability to give new KDE releases a try without waiting for 3 months for them. KDE PPA is awesome
- Easy upgrade every six months. An upgrade must work without backup/restore. I respect Clement's work on Mint very much, but for me, I don't want to backup and restore my system every six months.
- Repository for Skype, proprietary drivers, etc.
Have a laptop, and it is sometimes used by different people at my home. This means:
- Need encryption in case it gets stolen when I travel with it. Don't want full disc encryption, because that means typing two passwords. I want one single password.
- Don't want container-based encryption, because that means preallocating space for each user. I want something that grows dynamically, based on whatever a user needs. Ubuntu's approach for home encryption using individual files is just what I want. I know is not good for sparse files, but that's not my use-case.
- I have nvidia-optimus. So I need an easy way to install Bumbebee. Kubuntu gives you that. And nope, no BIOS switch so I need it in software
As far as I could tell, Kubuntu is the only distro that does what I need. So please don't let it die. Thanks
You think so? I just see vibrant colors, not 41megapixels detail. With a clear blue sky like that, and plenty of light, you minimize the problems of small sensors/lenses.
It is known that you can print up to A3 size with a 6 MP camera. So the only advantage would be when using digital zoom. If I zoom in those images I see a noisy sky. Sure, it can be a problem with jpg format. But in any case, the same picture could have been obtained with a 6MP camera.