As opposed to ODF with a bunch of backers with potentially unknown IP claims, admittedly incomplete spec and a patent grant only valid for versions of the specifications that are blessed by Sun Microsystems... which means that Sun gets all the say they want or there will be NO ODF revisions. Sun made an irrevocable IP covenant that it will not seek to enforce any of it's patents associated with ODF against any implementation of the specification. This is limited to versions of the specification in which Sun has had significant participation (read: not blessing, or even agreement, just participation). This is there so that Sun can keep new versions of the specification from including elements that infringe other Sun patents not covered by the initial covenant or to voluntarily include those additional patents in the covenant, not to keep the specification under Sun's control.
The stats are right there in the summary. OS X shares have not dropped significantly, certainly not enough to account for the increase in Vista's share. The only OS on the market with enough share to lose to give Vista > 6% is Windows XP, so that has to be where Vista is gaining from.
That's nonsense. The USA will not be able to stop proliferation among hostile countries by offering them a nuclear umbrella. Do you honestly think the Iranians would trust the USA to provide them with a nuclear deterrence? No, but it probably has stopped the Saudis from developing a nuclear weapons program to protect them from Iraq and Iran.
Your argument may hold for stopping proliferation among allied countries, but why should the USA care if allied countries develop nuclear weapons? They're called allies for a reason. To have any semblance of credibility when asking hostile nations to disarm, the USA will have to dismantle its own nuclear arsenal. Anything else is hypocritical, on the most basic level. Well of course it will only work on our allies, but when our allies in a region aren't developing nuclear weapons, that reduces their regional enemies' urgency to develop nuclear weapons. It doesn't stop proliferation, but it reduces the desire, and therefore makes it harder to justify the cost of proliferation, to our allies and adversaries alike. The only other way to prevent our adversaries from developing nuclear weapons is by force, which I gather you would be against.
I'm just curious if this is a security hole since I feel that if the directory tree is on by default that is a major security hole because I wouldn't want others to browse the site tree with files such as the one mentioned in the article. Good security generally requires that you do not put files containing passwords (especially clear text passwords) in the web server's document root. Turning off directory listing would not prevent the file from being accessible to anyone on the internet, it's just security through obscurity.
None of the 9-11 hijackers had any connection to Iraq, and Saddam didn't care for radical Shiite Islamic Fundamentalism! I minor detail, but the 9/11 hijackers were not shiite muslims.
If America wants to encourage countries not to proliferate, would it not make sense to disband our own arsenal? Absolutely not! One of the best tools we have to stopping proliferation is saying the USA will use its arsenal as a deterrent force so those countries will not need their own. That is why most European countries do not have their own nuclear weapons program, because during the Cold War we used our arsenal to extend the MAD principle to protect them.
Well yeah, but if you're working on a significantly complex application, you development, build and test environments shouldn't be on the same box, let alone on a laptop. I know that eclipse consumes RAM like it's going out of style, but tomcat is really pretty light, and your test databases should be held mostly on disk, it's not like you need great response times from your DB for testing if your code is ok. Outlook is a WTF there, but I understand it's required. Compiling I know can peg your CPU, but it'll do that regardless of CPU, it just might not take as long to finish on a faster one. Generally you shouldn't have to recompile your entire webapp, just the components you're changed, especially in Java. In your case, you should have a dedicated box for compilation and testing, at the very least. I'm currently running Eclipse, Firefox, Apache, Outlook and MS Word, Memory is pegged at just over 800M (150M of that is just for Firefox), CPU is generally low (except when compiling). I don't see why any mid-range laptop couldn't easily handle all of this.
Modern laptops have more than enough muscle to run office apps, with the exception being specialists such as programmers and video editors. Why would a programmer's laptop need more muscle than one that runs office apps? Video editing, yeah. Graphics editing, I can see. But programming?
Last--FF needs a master password set to be even remotely secure with regard to passwords, while Opera does not. This seems like a big hole. If Opera has encrypted your passwords, then it must have a copy of the decryption key stored somewhere in order to read them. It would seem that your program's author just didn't know where the key way, or it would have been able to read the Opera passwords too. Someone can correct me on this if I'm wrong (not a big Opera user), but to me it sounds like security through obscurity.
But that would at least require the user to enter their username and password before it can be stolen. I think the problem with Firefox and Safari is that they automatically populate those fields when the page loads, the user doesn't have a chance to _not_ enter that information.
DEB and RPM are failures because they attempt to solve the same problem and divide the software world in 2. So are the dozens of install shield products also failures for all trying to solve the same problem? Is Microsoft's MSI packages a failure for the same reason? Wouldn't having to have a separate installer for EVERY APPLICATION be a bigger failure than having just two?
Nvidias installer has to compile its own module if Nvidia doesn't have one for you, same with VMware. VMware has no other option, if they open source their effort would be totally ripped off from them and the company would likely die. Yeah, because nobody ever made money on open source, and there are no open-source virtualization products available already. VMWare can make money, even with an open-source kernel module.
The Linux kernel CAN run modules from other licenses, but it likely violates the license, the kernel even maintains a flag for when this happens so that support can refuse to help you if you are using a non GPL module. The Linux kernel can use any code that can be re-licensed as GPLv2. For example, BSD code can be added to the kernel.
X11 is a failure because people spend more time working around its ridiculously worthless architecture than actually getting things done, and no one wants to do the hard work to write a new system and hence we don't have a good one. So X11 is terrible, but nobody seems to think they cane write something better. And you see no contradiction in that idea? Perhaps next time you can go into a little more detail about what exactly is wrong with X, and which alternative is better and why.
Gnome is more like a thin client than a desktop, and since its being targeted for use as a desktop in place of things like windows, id say its nowhere close to being a competent choice. Do you even know what Gnome is? Or a thin client for that matter?
Libraries change all the time in Linux, every time I update a program in yast it wants to grab new libraries, are those programs all broken? I'm not suggesting it can be avoided....its just massive complexity for little gain since the program must be updated for the new library anyway, and it annoys the hell out of people who have no desire to put their programs in repositories. Wow, so the new version of your program uses features added to the new version of a library, and you are shocked that you have to upgrade your library? That is completely different than your complaint that an upgrade to the library requires an upgrade to your program (something I've never had happen). As someone has already mentioned, you can have multiple versions of a library installed at the same time if backwards compatibility is broken. But for minor updates like bug fixes or security fixes, you should never have to upgrade your application.
You missed the part where I said one choice IS 50 choices, that means one open source distro can easily serve 50 different uses because you can do whatever you want with it, even modify the code. And yet we have people modifying them and RELEASING 50 different slightly tweaked versions of the exact same thing. And most of the main distros are not serving different purposes, they are competing with each other save for the desktop/server split, which is an acceptable difference and one that usually does require significant differences. Then stick with Microsoft if you like having the choice of any color you want, so long as it's black. Seriously, if choice scares you, don't use Linux.
In Ubuntu, I double click the.deb file and it will check for dependencies, download and install any missing dependencies, then install the program. I'll give you a real-world example of when this is easier:
1.) On Windows I had to install a Cisco program to get VPN access to my company's network. I run the installer (an exe) and it fails because it needs the.Net 2.0 framework. So I download.Net 2.0 framework and run it's isntaller (another exe) and it requires Windows Installer 3.0. So I download Windows Installer 3.1 (because 3.0 didn't work), install it (another exe), install.Net 2.0 framework (exe), then install the Cisco program (exe).
2.) On Ubuntu, I wanted to install jEdit, a very nice programmer's editor written in Java. I download the.deb from their source forge website and run it. It informs me that it needs Sun Java 5 JRE and Sun Java 5 bin (and I think one other, but I don't remember), and asks me if I want to download and install those packages now. I say yes, it downloads a couple of other.deb packages from Ubuntu's servers, installs them, then installs jEdit.
but I make no apologies for Linux's failures, and there are many, specifically RPM/DEB, the slightly different distros, multiple desktops, worthless X11 etc. Care to give a reason why you find these technologies 'failures'? No, probably not, judging by the rest of your post.
The Linux distro people also seem to think including libraries with programs is a bad idea since you can update libraries individually for bugs, but in reality that just breaks the program which must be updated anyway, putting more work on the repo maintainer and making things more complex. If a bug fix to a library breaks your application, then your application sucks because it wasn't using the library properly, or the library sucks for introducing new bugs in a bug fix. Neither of these are caused by the way Linux shares libraries, and your suggestion only serves to keep a broken library from being fixed. That may be ok for bugs, but a terrible idea for security.
Effort required to package programs is also not the issue, you need to give a reason why we NEED 2 different package formats You don't need 2 different package formats, most distros settle on only one. You also don't need 2 different distros, one will likely provide you with everthing you need. The reason *we* need different distros is because the one that provides everything *you* want may not provide everything *I* want. If you want someone else to decided what *you* get, then I'll be happy to tell you which distro to use. But I don't want you telling me what *I* get.
open source means one choice IS 50 choices, since you can do what you want with the system and you can remove what you want. Yeah, ain't it great? The reason there are 50 choices is because there are at least 50+1 people now getting exactly what they want, otherwise there wouldn't be so many choices.
Having so many REAL distros also causes serious problems for software companies, for instance VMware server isn't in most repositories, so they have to include 50+ kernel modules with their package to even come close to having one to match each version of the running kernel in every distro. Waste of resources, directly stems from too many distros (and breaking the kernel module interface all the time). VMWare already knows how to solve this problem, the just choose not to. BTW, nVidia provides a single package for their binary kernel driver, so obviously it's possible.
The current Linux situation is such that to have any real success your drivers must be in the kernel tree and can't even use another open source license (ridiculous in many cases), and your software must be in each repository for each disto (also ridiculous and a massive waste of resources and developers time at every level). Wrong and wrong. The Linux kernel can accept code from multiple open source licenses, and you can host your own repositories for your packages. Many of the packages I use don't come from Ubuntu's repos, but from the author's own repo, and I still get all the benefits of apt with them.
Nvidia already got around the ridiculous GPL-only driver requirements, so that little experiment in forced licensing failed and now we have closed graphics drivers anyway, and it looks like everything else is being held together by these stupid package managers that every other operating system on the planet seems to do better without. I got my closed-source nvidia driver through my package manager, and it get updated whenever my kernel gets updated. As for other operating systems, Unix has Sys5, Windows has MSI, the BSD's I think have their own. Even Mac OSX has.app folders which are effectively application packages.
My point is that until all this actually results in mainstream software (particularly PC games) being widely available for Linux, it's all a moot argument. I'm just trying to guess why such software is not currently available for Linux, and the only real reason I can see is that the myriad distros of Linux are just too much hassle for mainstream developers. If you can give a more plausible explanation, by all means please do so. There are several reasons games aren't being actively developed for Linux, multiple distros is far from the top of that list. Here's a few off my head:
1.) Linux has a very small market share, and what share it does have is made up of people who don't generally like spending money on software, and many who don't generally have an interest in playing games on their PC.
2.) Most games companies have long ago invested in the Microsoft toolkits, like DirectX/Direct3D. This means that they not only have to port the games to Linux, they have to change the underlying technology they use. Games that were based on OpenGL have a much easier time moving to Linux.
3.) Video card manufacturers are still making crap drivers for Linux. Intel has the good open-source drivers, but they don't really make "gaming" cards. nVidia is alright most of the time, but still has some really annoying bugs that should have been fixed long ago. ATI is just a joke from everything I've heard. Open-source drivers from nVidia and ATI would do more to get games written for Linux than merging distros ever would.
so far, I've not encountered a commercial vendor for a particular software with the relevant licenses to legally allow the playing of.wmv/.aacs/.mp4 etc. I believe Linspire and Xandros have the relevant licenses for this. Additionally, Fluendo will be selling native linux codecs for these formats with the proper licensing.
Then perhaps you should legally license the codecs. There are several Linux distros that contain legally licensed codecs, specifically Linspire and Xandros come to mind. Are they free? No, of course not, because those codec licenses aren't free.
Except that's what this whole debate is about: if there was a "one-size-fits-all" Linux distro, then more software developers would develop for it, it would become more popular, and it would gain more market-share leaving Windows XP with considerably less than 75% of the market. Then the premise of the whole debate is flawed. It takes virtually no extra effort to make both an RPM and a DEB file when you already plan on making one or the other. This isn't *porting* you app to 2 different distros, this is *packaging* you app for 2 different package managers. It's like creating a.msi installer and a.exe installer for Windows.
And most of those same people have admitted that having a.rpm and.deb is not user-friendly enough for the average home PC user, meaning any software intended for a typical user that's currently on Windows will still have to be packaged specially for each distro and sub-distro. Again that's wrong. I've installed generic.deb files on Ubuntu 7.04, it far easier than installing anything on Windows. I've installed a generic.rpm file on Redhat, CentOS and Suse. Heck, I've even converted a generic.rpm into a.deb and installed it on Ubuntu.
Until you can literally package your software into a self-extracting archive that works like an.exe and works on all current distros and versions of linux as well as all GUIs this fragmentation is a problem. You can already do that, some people have. Notably, Sun's Java used to come in a binary installer for all Linux distros. Downsides to this approach are that you make the user resolve dependencies (dependency-hell), or package copies of the dependencies with your installer (dll-hell). You also have to write and run your own auto-updater if you want to provide automatic updates. But why should you need to write 2 or 3 programs? One being the program you want to provide, the others being an installer and update service? Especially when a perfectly good installer and update service is already on the system you're targeting. Even Microsoft recognized that this is the best approach, and has been pushing.msi installation packages for some time now.
I thought you were looking at consumers? 99% of them don't use specialist apps. Unfortunately, 99% of business users do run specialist apps. They don't need to, and usually there is an open source product that does everything they need, but 99% of them have been sold on a crap product by some vendor, and that product only runs on Windows. If they have only one crap product, then switching may be easier, but most have a half dozen crap products that only run on Windows, and getting them to switch all of them at once gets met with quite a bit of hostility.
Seriously, for most consumers, assuming Linux is still going after Windows and the desktop, more choice is not necessarily better, especially not when it numbers in the hundreds. Which is exactly why nobody buys a car, there are just too dang many to choose from, and from too many manufacturers. Ford alone has dozens of models! How can they expect consumers to make a choice between them? Until the automobile industry realizes this problem, the horse and buggy will remain king.
Slashdot Mirror
The stats are right there in the summary. OS X shares have not dropped significantly, certainly not enough to account for the increase in Vista's share. The only OS on the market with enough share to lose to give Vista > 6% is Windows XP, so that has to be where Vista is gaining from.
Thankfully I've never had to muck around in C/C++ programs of any size, and there is no comparable problem with Java.
Well yeah, but if you're working on a significantly complex application, you development, build and test environments shouldn't be on the same box, let alone on a laptop. I know that eclipse consumes RAM like it's going out of style, but tomcat is really pretty light, and your test databases should be held mostly on disk, it's not like you need great response times from your DB for testing if your code is ok. Outlook is a WTF there, but I understand it's required. Compiling I know can peg your CPU, but it'll do that regardless of CPU, it just might not take as long to finish on a faster one. Generally you shouldn't have to recompile your entire webapp, just the components you're changed, especially in Java. In your case, you should have a dedicated box for compilation and testing, at the very least. I'm currently running Eclipse, Firefox, Apache, Outlook and MS Word, Memory is pegged at just over 800M (150M of that is just for Firefox), CPU is generally low (except when compiling). I don't see why any mid-range laptop couldn't easily handle all of this.
Here's hoping that HTML 5 finally includes the tag.
But that would at least require the user to enter their username and password before it can be stolen. I think the problem with Firefox and Safari is that they automatically populate those fields when the page loads, the user doesn't have a chance to _not_ enter that information.
In Ubuntu, I double click the .deb file and it will check for dependencies, download and install any missing dependencies, then install the program. I'll give you a real-world example of when this is easier:
.Net 2.0 framework. So I download .Net 2.0 framework and run it's isntaller (another exe) and it requires Windows Installer 3.0. So I download Windows Installer 3.1 (because 3.0 didn't work), install it (another exe), install .Net 2.0 framework (exe), then install the Cisco program (exe).
.deb from their source forge website and run it. It informs me that it needs Sun Java 5 JRE and Sun Java 5 bin (and I think one other, but I don't remember), and asks me if I want to download and install those packages now. I say yes, it downloads a couple of other .deb packages from Ubuntu's servers, installs them, then installs jEdit.
1.) On Windows I had to install a Cisco program to get VPN access to my company's network. I run the installer (an exe) and it fails because it needs the
2.) On Ubuntu, I wanted to install jEdit, a very nice programmer's editor written in Java. I download the
aptitude is just a front end to apt, just like synaptic and adept are just front ends to apt. All the magic still happens in apt.
Yes, a wonder spell call 'apt'.
Canonical doesn't write any of those programs either, and yet Ubuntu automatically updates them for me anyway.
1.) Linux has a very small market share, and what share it does have is made up of people who don't generally like spending money on software, and many who don't generally have an interest in playing games on their PC.
2.) Most games companies have long ago invested in the Microsoft toolkits, like DirectX/Direct3D. This means that they not only have to port the games to Linux, they have to change the underlying technology they use. Games that were based on OpenGL have a much easier time moving to Linux.
3.) Video card manufacturers are still making crap drivers for Linux. Intel has the good open-source drivers, but they don't really make "gaming" cards. nVidia is alright most of the time, but still has some really annoying bugs that should have been fixed long ago. ATI is just a joke from everything I've heard. Open-source drivers from nVidia and ATI would do more to get games written for Linux than merging distros ever would.
Then perhaps you should legally license the codecs. There are several Linux distros that contain legally licensed codecs, specifically Linspire and Xandros come to mind. Are they free? No, of course not, because those codec licenses aren't free.