What Are The Biggest Problems With Linux?

[fret]

Face it. Linux is not a commercial product. It was never meant to amass vast amount of money from clueless people as its sole purpose. Linux doesn't have to cater to the mass. There is no reason to.

It was created to scratch the itch that people had in the computing world. And they formed communities who had similar itches so they can scratch common itches together.

The biggest problem is the people who have problems with Linux. If they have an itch, don't complain about Linux. Form a community to solve the problem. Pay someone to solve the problem.

Linux is not a free handout for lazy people who expect hand holding in every step just like in commercial products. The "free" in Linux is "freedom", not "gratis". Freedom to tinker with it to solve problems. Then share the solution with others, so others can improve it and give the improvement back.

It has to come from the people. If people want freedom in software, work for it, but don't treat it as an entitlement.

The next biggest problems would be software patents and shady monopolistic practices that hinder innovation. But again, these are controlled by those who would like to amass revenues from lazy clueless people who don't care about true freedom.

So back to the biggest problem: The people who have problems with Linux and expect free solutions without contributing anything.

[WorBlux]

Face it. Linux is not a commercial product...

No, it is a commercial product, look at all the embedded, server, and high-performance systems that it operates on every day. Big money, and they tend to pay to get what they want.

It is not a consumer product, not really cut for the unwashed masses. Certain things that it would need to be are lacking and just need cold hard cash. There are a lot of talented developers interested, it's just that the need to feed a family or pay the mortgage.

[plonoma]

Could you elaborate on that please?
I'm very curious to what you are talking about.

Let's take that extra libraries for compatibility would cost us a gigabyte.
How does it compares percentage wise?

I'm talking that it should be possible to do this, not that everybody is forced to do this.

If you don't want bloat, fine, then don't include lots of libraries in your distribution.
And be forced to make sure programs work with the included version.

[gamerk2]

Why should 90% of Linux devices have to waste a lot of time cutting out bloat because a few percent want it?

"Bloat" is and always will be subjective. In the days of mutli-terrabyte HDD's, a few hundred megabytes of extra hardware support is minimal overhead, especially if said code is never actually run.

I find it ironic that some people love linux because of how easy it is to configure, then complain about not wanting to carry some features that enable linux to run on a wider variety of hardware.

Of course it does. It pushes your maintenance costs off onto the OS developers, who have to maintain crusty old crap forever.

Which is why Windows is so full of bloat, security holes and bugs. Microsoft even have to support old, undocumented bugs because SuperWizzoWriter 2000 crashes if they fix them.

I'd stop complaining about MS's security holes, since Windows is significantly better off then most other OS's on the market right now.

I think you miss the point.

Most Linux users don't care, and most Linux developers are building custom systems for Linux. We don't need to support crusty old APIs because we don't care that you don't want to invest any time in maintenance work to deal with API changes.

There are only two non-game, non-free applications I run on my Linux box and one of them has a Linux port; I prefer running the Windows version in Wine because I don't trust random software installers with root permission.

Translation: You don't run it, so you couldn't care if support was dropped for it.

[Yfrwlf]

The main problem in my opinion is the lack of standardization, not only between distributions, but also between desktop environments.
And Linux needs a common way to install/remove applications.

+99999999999999999

1. Developers, stop releasing source-only packages. Linux is never ever going to get mainstream adoption without functional, distro-agnostic binaries being made available for the major projects. This includes desktop environments. Stop forcing users to beg a distro company for updates. Linux is supposed to mean freedom. My first and foremost goal developing a project is getting it easily to users, but it seems many developers don't give a crap about users, ease-of-use, and installation.
2. Developers need to use a package standard which will get dependencies no matter what. I don't care if libjdwrujbs-2.125 isn't in my distro's repo, download that version from the library's website. Otherwise, if the library isn't nicely and readily available, get them to be a better project host, or include the specific version along with your program.
3. Use libraries which have better standards and don't constantly break the API so users won't need a different version for every program.

So far the only real solutions that seem to exist that I've found (feel free to give me better ones) is packaging systems like Zero Install which allows nice installation, updates, etc in a cross-distro fashion, so you install a 2nd package manager side-by-side with the existing one, and it plays nice. Also, using cross-desktop environment standards, like wxWidgets? Or can Qt also mimick GTK? You need to use something like that so your program looks good no matter the desktop environment the user is running.

If those are the best standards that Linux has right now, freedesktop.org and other standards bodies need to recommend them. It would be really great if Linux stopped being stupid with drivers too though, yes. Make a driver communication standard that never or very rarely has to change. If that means "making the Linux kernel sort of into a macro kernel" then why not? The reasons for not doing it are stupid. Making it easier to maintain and create a driver for Linux means win for Linux and its users.

[cdobrich]

Linux is a monolithic kernel and every time they update it it can break many things. That's why micro-kernels are better and device drivers should be in user space not kernel space.

You make a good point and perhaps your experience is different than mine, but from my perspective the breakage is minimal and easily mitigated by a dedicated distribution that is watchful of what it pushes out to its users.

And if you ever complain about missing hardware drivers, try using a micro-kernel and you'll really understand what no harddrive support means. Great example is throwing Windows XP onto a computer and it doesn't support the network card because it only comes with 31-flavors of network drivers, and the one I've got doesn't exactly fit. It's almost never the case with Linux there. Sure, you might still miss hardware, but you'll have a whole heaping lot more with a monolithic kernel. Not to mention Micro-kernels can (but not necessarily) be inefficient.

[bwat47]

+99999999999999999

1. Developers, stop releasing source-only packages. Linux is never ever going to get mainstream adoption without functional, distro-agnostic binaries being made available for the major projects. This includes desktop environments. Stop forcing users to beg a distro company for updates. Linux is supposed to mean freedom. My first and foremost goal developing a project is getting it easily to users, but it seems many developers don't give a crap about users, ease-of-use, and installation.
2. Developers need to use a package standard which will get dependencies no matter what. I don't care if libjdwrujbs-2.125 isn't in my distro's repo, download that version from the library's website. Otherwise, if the library isn't nicely and readily available, get them to be a better project host, or include the specific version along with your program.
3. Use libraries which have better standards and don't constantly break the API so users won't need a different version for every program.

So far the only real solutions that seem to exist that I've found (feel free to give me better ones) is packaging systems like Zero Install which allows nice installation, updates, etc in a cross-distro fashion, so you install a 2nd package manager side-by-side with the existing one, and it plays nice. Also, using cross-desktop environment standards, like wxWidgets? Or can Qt also mimick GTK? You need to use something like that so your program looks good no matter the desktop environment the user is running.

If those are the best standards that Linux has right now, freedesktop.org and other standards bodies need to recommend them. It would be really great if Linux stopped being stupid with drivers too though, yes. Make a driver communication standard that never or very rarely has to change. If that means "making the Linux kernel sort of into a macro kernel" then why not? The reasons for not doing it are stupid. Making it easier to maintain and create a driver for Linux means win for Linux and its users.

QT can integrate with GTK(2)themes. QT is a pretty nice cross platform toolkit, it looks decent in gnome and windows.

[Yfrwlf]

You make a good point and perhaps your experience is different than mine, but from my perspective the breakage is minimal and easily mitigated by a dedicated distribution that is watchful of what it pushes out to its users.

And if you ever complain about missing hardware drivers, try using a micro-kernel and you'll really understand what no harddrive support means. Great example is throwing Windows XP onto a computer and it doesn't support the network card because it only comes with 31-flavors of network drivers, and the one I've got doesn't exactly fit. It's almost never the case with Linux there. Sure, you might still miss hardware, but you'll have a whole heaping lot more with a monolithic kernel. Not to mention Micro-kernels can (but not necessarily) be inefficient.

You shouldn't have to depend on anyone. Freedom means anyone, even users who only know how to point and click, can install drivers. When modifications are required for every tiny version difference, that makes that an impossibility unless you create an automated compilation system. Having a standardized driver ABI is much more efficient and a nicer feature, though.

As for your "not having the right driver" Windows XP example, that is BS. Windows XP not coming with enough drivers on the installation medium vs. Linux installation mediums containing more drivers has nothing to do with anything except that open source licensing means all the drivers can legally be included, whereas Microsoft relied on proprietary drivers being supplied from outside sources. A Linux installation medium can contain the same number of drivers regardless of whether the kernel is micro or monolithic.

[Yfrwlf]

QT can integrate with GTK(2)themes. QT is a pretty nice cross platform toolkit, it looks decent in gnome and windows.

Since it's now open source, it might be a good standard to use then for Linux GUI apps. I don't know if wxWidgets has more features than Qt but I assume the Qt has a lot more work put into it and is probably more featureful of the two.

[asdx]

QT can integrate with GTK(2)themes. QT is a pretty nice cross platform toolkit, it looks decent in gnome and windows.

Qt, not QT.