Is Linux ready yet for typical home desktop users?

Page 3 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

DasFox

Diamond Member
Sep 4, 2003
4,668
46
91
Originally posted by: Nothinman
As long as they can call for help no problem, but guess what? Who are they going to call and get FREE Linux tech support is what I'd like to know?


That hasn't been a real problem for years. Pretty much any piece of software worth running is available in a repository that apt or yum can use and as such will be able to handle any dependency resolution automatically.

Sorry if I wasn't clear here, I don't mean meeting the actual issue of dependencies for a install, but many times something you installed changed a version of something, and then the dependancy for this towards another program changes and has problems.

As example, a dependant that comes to mind; GTK, you get a update, that requires a update to GTK, for this particular app, but then other apps needed a different version of GTK, things like that sometimes can get messy.

But it's not true, I can't remember the last time an update broke anything for me, even the XFree86->Xorg and Xorg6.9->Xorg7.0 transitions were painless. If that really does happen, it's a bug in your distribution and you need to tell them about it.

Its is true, this has nothing to do with the distro, unless the team from that distro, recompiled, patched and built things, then yes I've seen issues having problems this way.

I'm saying also besides this, when new versions of things come along, many times your hardware support will change, and things aren't working anymore, or like they use to, and a software application that wanted a particular version of something, just doesn't seem to work, or run as well on a newer version. I just called this breaking, if it's really the correct way to say, but many others in Linux will call it that, especially when something stops working.

I use to run a Slackware support site, and I have been compiling software for 7 years as well. I do know for a fact developers will build a app for just a particular version of a dependancy, and if you try to use another version, things just don't work as they should, or at all.

Now granted when we are talking about what I would call the big names in dependancies, then going from version to versions shouldn't make for many problems, but at times they do.

This happens all the time with people in Linux, I have been active in the Linux community for the past 7 years as I mentioned, I know this for a fact first hand.

Go into any hardcore IRC Unix/Linux server so you can chat, and see Linux users, and you too will see the masses of people having issues just like these on a daily basis.

-----> irc.freenode.net ;)
 

Nothinman

Elite Member
Sep 14, 2001
30,672
0
0
Sorry if I wasn't clear here, I don't mean meeting the actual issue of dependencies for a install, but many times something you installed changed a version of something, and then the dependancy for this towards another program changes and has problems.

Again, in a well designed system that's not a problem as the new version of X won't be installable until Y is available too. If the libraries themselves changed and introduced a compatibility problem, you can't avoid that even on Windows.

As example, a dependant that comes to mind; GTK, you get a update, that requires a update to GTK, for this particular app, but then other apps needed a different version of GTK, things like that sometimes can get messy.

Oh please tell me when that's actually happened to you. 99% of my apps are GTK and I have never had a problem with dependencies, incompatibilities or anything relating to library changes. There's no doubt it's possible, but it's also possible that you'll die in a fiery car crash tomorrow and that doesn't stop you from leaving your house, right?

Its is true, this has nothing to do with the distro, unless the team from that distro, recompiled, patched and built things, then yes I've seen issues having problems this way.

That is the point of a distribution, they maintain the packages for you. They build and test them before they're uploaded to the servers for everyone else to use and most have an intermediary server too so that more people can test them before they're considered good and pushed out to everyone else.

I'm saying also besides this, when new versions of things come along, many times your hardware support will change, and things aren't working anymore, or like they use to, and a software application that wanted a particular version of something, just doesn't seem to work, or run as well on a newer version.

Once again, IME that almost never happens. That's called a regression and if it does happen it's a bug and should be fixed before the package makes it to the end users. The only thing that's even remotely problematic is the closed source nVidia drivers and that's because I go off on my own and don't use the prebuilt packages in my distribution, if I did I wouldn't have to worry about them either.

I use to run a Slackware support site, and I have been compiling software for 7 years as well. I do know for a fact developers will build a app for just a particular version of a dependancy, and if you try to use another version, things just don't work as they should, or at all.

Then I would say that things have changed since you ran that site, either that or those particular libraries were crap and probably aren't being used any more.

Now granted when we are talking about what I would call the big names in dependancies, then going from version to versions shouldn't make for many problems, but at times they do.

Actually I would say the big names have the biggest chance for problems, especially things like the XFree86->Xorg transition and just about any current Linux kernel update since development is moving so fast. And actually udev has been a bit of a problem child, but Linus has banged it into gregkh's head that backwards compatibility is important and I doubt udev will be much of a problem any more. But because the Debian X maintainers took their time and did it right, I barely noticed the transition to Xorg. Infact I had to go look at the package versions to believe that it had happened. And that's how it's supposed to work, it's nice to have competent people behind the scenes.

Go into any hardcore IRC Unix/Linux server so you can chat, and see Linux users, and you too will see the masses of people having issues just like these on a daily basis.

I do idle on IRC still, but I don't pay much attention to many channels besides #linux on the Ars server any more. And I'm even on freenode, but only in the #suspend2 channel.
 

DasFox

Diamond Member
Sep 4, 2003
4,668
46
91
Again, in a well designed system that's not a problem as the new version of X won't be installable until Y is available too. If the libraries themselves changed and introduced a compatibility problem, you can't avoid that even on Windows.

Maybe I'm not making myself clear, you sound like a pretty experienced Linux user. From what you are saying, yes at the LEVEL of the OS, things aren't typically going a muck, but at times they do.

At the small application level too I have seen this happen all the time, no offense but it doesn't sound like you are using that much software in Linux.

Looks lets make sure we are talking about and including ALL software from A-Z that is available for Linux. I'm not just talking the major mainstream here, but any and everything that is out there for Linux.

Oh please tell me when that's actually happened to you. 99% of my apps are GTK and I have never had a problem with dependencies, incompatibilities or anything relating to library changes. There's no doubt it's possible, but it's also possible that you'll die in a fiery car crash tomorrow and that doesn't stop you from leaving your house, right?

I can tell you just last month I had problems with dependancies in Gentoo, dealing with Acidrip.


That is the point of a distribution, they maintain the packages for you. They build and test them before they're uploaded to the servers for everyone else to use and most have an intermediary server too so that more people can test them before they're considered good and pushed out to everyone else.

Ok WAIT, hehe this is where we fouled up in this conversation, not every distro out there can maintain every piece of software available for Linux, some do maintain more then others, but even the big names sometimes don't even keep up with some names that are even considered big. So then end-users have to go out there and find like a rpm from somewhere else and then install this.


But I have seen time and time again packages maintained by even distros as big as Ubuntu that ran like crap.

Did you know, and they might still even be doing it, that Ubuntu use to compile all their package for the Pentium CPU. You know what that means for people that use AMD cpus?

Really poor performance, example I have a AMD XP 3000+, now fast enough for Mplayer, but because Ubuntu did this, Mplayer would not run at all for me.

Also I remember last year MPlayer and Transcode was lacking support in Ubuntu, meaning to get things working for it, you had to go outside the "Stable" tree to download it and use them, oh no more issues here with unstable packages.

Once again, IME that almost never happens. That's called a regression and if it does happen it's a bug and should be fixed before the package makes it to the end users. The only thing that's even remotely problematic is the closed source nVidia drivers and that's because I go off on my own and don't use the prebuilt packages in my distribution, if I did I wouldn't have to worry about them either.

We are talking bugs here for many things, hardware support, compile flags compiled into the software for certain functions and hardware compatibility.

No developer can build a program to encounter all the likely situations with hardware, software, etc.. that one runs on their box. This is the reality of software on any operating system.

Then I would say that things have changed since you ran that site, either that or those particular libraries were crap and probably aren't being used any more.

Nothing has changed, bugs are bugs and as long as people code mistakes and are not perfect we will have problems. What makes you think we live in a bug free world in Linux?

Actually I would say the big names have the biggest chance for problems, especially things like the XFree86->Xorg transition and just about any current Linux kernel update since development is moving so fast. And actually udev has been a bit of a problem child, but Linus has banged it into gregkh's head that backwards compatibility is important and I doubt udev will be much of a problem any more. But because the Debian X maintainers took their time and did it right, I barely noticed the transition to Xorg. Infact I had to go look at the package versions to believe that it had happened. And that's how it's supposed to work, it's nice to have competent people behind the scenes.

Big or little it doesn't matter they all face issues. I've seen it from the biggest to the smallest.

Look I'm not pulling your leg here, I've used and use Linux as a profession and my level of experience is extremely high.

I think you might be looking at this from one side, because all the things I am saying are correct and happen to people on a daily basis.

If you are a Linux user and lover of the OS, you should hang out in some of the big channels on freenode, #debian, #gentoo, #ubuntu, just to name a few and see all the problems people face that have I been saying.

I know for a fact because I have dealt with this as a software packager for Slackware for 7 years.

ALOHA

For the sake of all arguments, you and many others might not ever face, or hardly see problems, but then there are those that will, this is a fact.

So for all those that don't have problems, guess again, there are that many more that do, and always will, that's why they are called bugs, some get them, some don't.

I always seem to have to explain this concept of bugs to people. ;)


ALOHA
 

Nothinman

Elite Member
Sep 14, 2001
30,672
0
0
At the small application level too I have seen this happen all the time, no offense but it doesn't sound like you are using that much software in Linux.

Depends on what you consider a lot, but I do everything on my computers on Linux.

Looks lets make sure we are talking about and including ALL software from A-Z that is available for Linux. I'm not just talking the major mainstream here, but any and everything that is out there for Linux.

It's impossible to include all software ever available for Linux.

I can tell you just last month I had problems with dependancies in Gentoo, dealing with Acidrip.

Then I would blame Gentoo, they're not exactly known for their QA process anyway. And I have used acidrip, not on a daily basis or anything but I've never had it break when I wanted to use it.

Ok WAIT, hehe this is where we fouled up in this conversation, not every distro out there can maintain every piece of software available for Linux, some do maintain more then others, but even the big names sometimes don't even keep up with some names that are even considered big. So then end-users have to go out there and find like a rpm from somewhere else and then install this.

I know I'm biased, but I would say that in the great majority of cases if the software isn't packaged in Debian (and thus Ubuntu with universe and multiverse) it's probably not worth using. It's usually much better and simpler to just find another piece of software that is packaged and does the same job.

But I have seen time and time again packages maintained by even distros as big as Ubuntu that ran like crap.

Of course it's possible. One thing to consider with Ubuntu is that even though Ubuntu is big in name they're not in man power, especially compared to Debian, and each package doesn't have a specific maintainer like they do in Debian, so it's possible that someone who doesn't know the package as well as the Debian maintainer has uploaded the Ubuntu package. But once again, all you have to do is file a bug report. It's not like it's a routine occurance.

Did you know, and they might still even be doing it, that Ubuntu use to compile all their package for the Pentium CPU. You know what that means for people that use AMD cpus?

Really poor performance, example I have a AMD XP 3000+, now fast enough for Mplayer, but because Ubuntu did this, Mplayer would not run at all for me.

I find that hard to believe because I've been using mplayer on a set of 1.2Ghz TBirds for years and have never had any performance issues. The only time I've seen it run slowly is when I mistakenly told it to use X11 output instead of Xv. But mplayer is a very special case, for a while now they've had runtime CPU detection but it was (or at least was perceived to be) problematic so it was disabled by Marillat in his Debian packages. A while back, not sure how long but it's been at least a few months, all of the CPU-specific builds were disabled and the runtime detection enabled.

And on top of that mplayer isn't really supported by Ubuntu, it's in multiverse. Their main focus for a video player is totem and the gstreamer libraries, which is what any new user would have been using.

Also I remember last year MPlayer and Transcode was lacking support in Ubuntu, meaning to get things working for it, you had to go outside the "Stable" tree to download it and use them, oh no more issues here with unstable packages.

Of course you have to go outside of main to get them, hell they're not even in Debian because of possible patent violations in the software. But once again, I really doubt anyone just starting with Linux will be specifically looking to run mplayer and transcode unless told to do so by some more experienced user.

We are talking bugs here for many things, hardware support, compile flags compiled into the software for certain functions and hardware compatibility.

Hardware support in 99% of the applications out there is non-existant, as long as the driver in the kernel works the application will too. And if you don't run Gentoo you won't have to worry about problems with your compiler flags, the maintainers who understand the software know what's safe and what isn't and generally as long as you don't do anything stupid trying to optimize the build to hell you won't have any problems.

No developer can build a program to encounter all the likely situations with hardware, software, etc.. that one runs on their box. This is the reality of software on any operating system.

That's why the fact that Debian supports nearly a dozen architectures and has autobuilders for them all is such a big deal, it catches errors that developers with access to only 1 or 2 architectures wouldn't catch. And there are a decent amount of Debian users running 'odd' architectures like sparc, Alpha and IA64, so while not everything can be tested 100% they do catch most of the problems up front.

Nothing has changed, bugs are bugs and as long as people code mistakes and are not perfect we will have problems. What makes you think we live in a bug free world in Linux?

I'm not saying it's bug free, that can be easily disproven by looking at bugs.debian.org. But I can't remember the last time I ran into a bug that caused me any significant problems. I can think of one cosmetic bug in Nautilus and one in Galeon's tab handling, but nothing that affects functionality.

I know for a fact because I have dealt with this as a software packager for Slackware for 7 years.

Ah, a slackware packager, that explains a lot =)

For the sake of all arguments, you and many others might not ever face, or hardly see problems, but then there are those that will, this is a fact.

So for all those that don't have problems, guess again, there are that many more that do, and always will, that's why they are called bugs, some get them, some don't.

No doubt, but I would wager that they would experience an equal or higher number of problems/bugs in the Windows world. The difference is that the bugs on a Linux system are usually much easier to track down and fix, hell I've been involved with a support call to MS for the past 2 days and not even MS can tell us why one of their products is acting the way it is. And on top of it they've already mentioned to us one bug in their documentation, it states that a setting in one of their ini files is ignored when in reality it's not, it's just used for another piece of the program.
 

drag

Elite Member
Jul 4, 2002
8,708
0
0
How is compiling mplayer so that it works on a 586 proccessor make it to slow to be used on a AMD k7 cpu? I never had any problems. Weither mplayer was compiled for 486 or 586 or 686 or whatever it didn't enable me to all of a sudden play movies I couldn't before.. unless there was some other bug going on.

Do you think that Microsoft compiles optimized versions of their mplayer for Windows?

The dirty secret behind all these march="arch" stuff is that they don't actually do much of anything on the x86 platform. There are a few things that make very modest improvements to certain applications that are programmed to take advantage of mmx or sse instructions, but that's about it.
 

DasFox

Diamond Member
Sep 4, 2003
4,668
46
91
I hear ya Nothinman ;)

drag, compiling march or mcpu as, 486, 586, 686 is generic, the problem is when you get specific with doing them as march=pentium4 or mcpu=pentum4 as example and this what Ubuntu use to do before.

Their software was compiled exclusively for Pentium and when using a AMD in certain situations this can cause problems.

ALOHA
 

Nothinman

Elite Member
Sep 14, 2001
30,672
0
0
Their software was compiled exclusively for Pentium and when using a AMD in certain situations this can cause problems.

Personally, I'd say that's a bug in the hardware then.
 

drag

Elite Member
Jul 4, 2002
8,708
0
0
Originally posted by: DasFox
I hear ya Nothinman ;)

drag, compiling march or mcpu as, 486, 586, 686 is generic, the problem is when you get specific with doing them as march=pentium4 or mcpu=pentum4 as example and this what Ubuntu use to do before.

Their software was compiled exclusively for Pentium and when using a AMD in certain situations this can cause problems.

ALOHA


So your telling me they sent out a mplayer package that was compiled for march for Pentium4?!

I know that when using 3rd party packages for Debian you'll may have a mplayer stand as a virtual package for various optimized mplayer packages.. Like for my laptop I have mplayer-g4 or whatnot. Are you sure that's not what happenned?

I thought when you said they optimized for 'pentium' then that's Pentium 1 and should work with AMD cpus with no problems whatsoever. However with march=pentium4 that will cause serious problems.

If Ubuntu installed mplayer-p4 or whatnot by default when you tried out install mplayer then that's just realy realy stupid.
 

Nothinman

Elite Member
Sep 14, 2001
30,672
0
0
I know that when using 3rd party packages for Debian you'll may have a mplayer stand as a virtual package for various optimized mplayer packages.. Like for my laptop I have mplayer-g4 or whatnot. Are you sure that's not what happenned?

Apparently you haven't looked at them in awhile, all of the mplayer-<arch> packages are dummy packages now.
 

drag

Elite Member
Jul 4, 2002
8,708
0
0
nope. Haven't looked in a while. Your right.

I just use the software, I don't pay much attention to it after I got it installed. I just update everything about every other week or so. :)
 

DasFox

Diamond Member
Sep 4, 2003
4,668
46
91
Well they did do mcpu=pentum4 in the past, which is really stupid for the AMD arch, but I think they saw the light now, at least from Nothinman's saying with these dummy packages now.

Actually I remember it now, they compiled all of Ubuntu in the past as:

march=i586 mcpu=pentium4 :(
 

scottws

Senior member
Oct 29, 2002
468
0
0
Originally posted by: Nothinman
As example, a dependant that comes to mind; GTK, you get a update, that requires a update to GTK, for this particular app, but then other apps needed a different version of GTK, things like that sometimes can get messy.

Oh please tell me when that's actually happened to you. 99% of my apps are GTK and I have never had a problem with dependencies, incompatibilities or anything relating to library changes. There's no doubt it's possible, but it's also possible that you'll die in a fiery car crash tomorrow and that doesn't stop you from leaving your house, right?
Fairly recently, GAIM 2.0 beta 2 required a newer version of the GTK+ library, but GIMP required an older version. At the time, it was not possible to run both GAIM 2.0 b2 and GIMP due to library incompatibility.

But this was on Windows. I'm not sure if the same problem existed on Linux for these programs and libraries or not.

But that is a very recent example of two pretty big software programs where there was a library incompatibility.
 

Nothinman

Elite Member
Sep 14, 2001
30,672
0
0
But that is a very recent example of two pretty big software programs where there was a library incompatibility.

And you wouldn't have seen that on a decent Linux distribution because the Gaim and Gimp versions would have had to work with the same version of GTK2. This might have meant that you couldn't have gotten Gaim B2 from the repository, but from the sounds of it that would be preferable for most people anyway.
 

Modular

Diamond Member
Jul 1, 2005
5,027
67
91
Originally posted by: bersl2
A typical one? No. Then again, IMO, typical home desktop users shouldn't even be using a general purpose computer in the first place.

Of course, desktop Linux doesn't need the typical user. It needs the hardware support. FOSS drivers can come with time, as long as devs don't become complacent about this eventual goal.


Saying that the "typical" desktop user shouldn't use a PC in the first place isn't fair...I understand the frustration with the idiocy of the "average user" but saying they shouldn't use a PC is like saying that you shouldn't drive a car if you don't know how it works, or you shouldn't use electricity because you aren't an expert in elctron flow and nuclear technology...

The home PC is such a mainstream NECESSITY that no matter which way you slice it, it needs to be user friendly: pretty much every mainstream app/hardware peice should be pretty much plug and play.

People who use Linux use it for that very reason, sometimes it's a challenge. They like to mess with drivers and hardware to make it work. I'm stepping into that world now...but until Linux becomes more user friendly there is no way it can ever be mainstream...

 

Nothinman

Elite Member
Sep 14, 2001
30,672
0
0
Saying that the "typical" desktop user shouldn't use a PC in the first place isn't fair...I understand the frustration with the idiocy of the "average user" but saying they shouldn't use a PC is like saying that you shouldn't drive a car if you don't know how it works, or you shouldn't use electricity because you aren't an expert in elctron flow and nuclear technology...

That's a terrible analogy, it takes no electrical knowledge to plug an AC adapter into an outlet and I don't need to know how the engine works to drive the car. Computer software just isn't that abstracted from the underlying system yet and it won't be as long as the boxes have to be general purpose, do-anything type boxes. And the biggest difference is that most people know that they don't understand their house's electrical system or their car's engine so when something needs fixed/changed they call someone who does know. But with computers everyone thinks they're an expert so they poke around and make it worse.

The home PC is such a mainstream NECESSITY that no matter which way you slice it

Why? People have been living without them for centuries and many still do, they're a really nice convenience but they're not anywhere near a necessity yet.

People who use Linux use it for that very reason, sometimes it's a challenge. They like to mess with drivers and hardware to make it work. I'm stepping into that world now...but until Linux becomes more user friendly there is no way it can ever be mainstream...

No, I use it because I don't like to mess with that crap and 99% of the time my Linux boxes 'just work'.