Operating Systems > Linux and UNIX

What needs to be done about Linux

(1/9) > >>

slave:
I'll be the first to admit how much I love and use Linux.  It's what got me really interested in computers in the first place.  The problem is that it's not satisfying to me for only nerds or tech-oriented people to use it.  I want everyone to be able to easily use Linux, no matter what their level of technical expertise is.  Actually, Linux is very easy - if you are used to its non-intuitive and often archaic configuration systems.  Almost any "Linux geek" learned Linux by adapting his mindset to Linux - by thinking like the operating system.  I want Linux to be better than that.  Computers should adapt to the users - not vice-versa.  I want Linux to be for everyone, and I want it to have an interface that anyone can grasp.  It should be as easy to use as a Mac.  This does not mean I want Linux to become an OS that technical people can't use and love.  That is the last thing I want to do.  Rather, I would like to see the Linux desktop be easy, intuitive, and simple, while allowing the more technically inclined of us to easily delve into the things we know and love about linux like the shell, configuring via text files, compiling programs, tweaking, installing 6 different browsers, etc.  And as I technical user, I appreciate the artistic and intuitive nuances that went into creating OS X.  I think it is a myth that all technical users are uninterested in having an intuitive graphical interface.  They simply have learned to live without it - the negative side to this is they often expect everyone else to live without it too.  

I believe Linux can be made to be the operating system of choice for nontechnical users, but things need to be done in order for that to happen.  So I'm going to do something bold and list all the major blockers Linux has that keep it from being used everyday by the bulk of computer users, and I will try to suggest solutions to these problems that are realistic.


Problem #1 - The explosive pace of Linux development makes it difficult for end-users to stay put.

Linux distributions release too often.  2-3 times a year? Give me a break.  Most people, unless they are like me and have some sort of obsessive disorder, don't enjoy installing operating systems.  We need a Linux distribution that focuses on getting it right instead of simply spontaneously releasing another OS update just because it's been 5 months since the last update.  The Debian release model should set a good example.  Debian only releases, to quote, "when it's done", which is usually around every two years.  This is why Debian has traditionally been one of the most solid operating systems to use in a production environment.  If there was a desktop Linux distribution that followed this model, it would allow for more stability and would also allow companies producing software that runs on Linux to more easily assure users that the software will be compatible.  It's much easier to handle library dependencies and customer support when your OS releases every 2 years instead of two times a year.

Also, when  the OS did release updates, they would be all that more dramatic.  ;)

Problem #2 - No easy way to install programs, especially 3rd party programs.

Well, this is a fairly small problem actually, because we have apt-get, which makes managing programs on Linux rather painless, but it has a problem.  There are no simplified front-ends to apt-get. Synaptic doesn't count; it's still too complicated for the average man.

What we need is a nice frontend to apt-get and dpkg (I am assuming here that this distribution will be based on Debian)  that behaves like installers on Mac or Windows.  The question is which one?  In Windows you usually are presented with a "Wizard" which guides you through installation of the program using a dialoge-based walkthrough, typically asking you a bunch of questions users shouldn't need to answer, like where do you want it installed and if you want to view the Readme file (like anyone clicks that)  I suggest a front-end where users can simply drag a program icon to a "programs" folder and have it install the program automatically, and download the dependencies if needed - although if the package was designed for this distribution then that shouldn't be necessary.  This "meta-folder", so to speak, would only contain applications, not libraries and trivial utilities that you see all over the Synaptic menu.

Problem #3 - hardware detection and other issues.

Theoretically, hardware detection on Linux is actually quite good.  The little hardware that isn't supported is almost always due to hardware companies being difficult and refusing to release the specs of their chipsets.  This, I am confident, will change in the future as Linux becomes more and more important.

I said "theoretically" because most distributions do a half-assed implementation of hardware detection and management. Even the mainstream distros like Fedora don't seem to do it right.  For instance the Red hat hardware detector, kudzu, lets you know only on system startup if there is a device added, and comes up with an ugly menu to tell you this, which you have to press a button or it will do nothing.  Nobody should have to press a button.  That's not what "Plug and Play" is all about.  It should be like plugging a Gamecube controller in.   This is certainly possible on Linux, and it could be implemented in a superior way than Windows currently does it.  In Windows, when you plug in a new device, a crummy bubble notification comes on and alerts you of this.  This is intrusive and unnecessary.  The ideal thing would be for it to only come up with a window or notification if it saw you plugged in something but it couldn't find the correct drivers for it.  This may be the case in the future when more proprietary drivers are made for Linux than are today.  Since I want this distribution to be entirely composed of free software, it would be unacceptable to include these drivers.  But there could be an apt repository included that would let you install a proprietary driver if you so wanted to.

Problem #4 - X window system

I will not say much about X here, but I would like it to get the same abilities that quartz has.  This is really just an issue of time, so there isn't much to be said, really.

Problem #5 - General user interface

KDE and GNOME provide a decent user interface, but they could be better.  I know the goal of the projects is to be portable, so having Linux-specific configuration tools isn't something being worked on currently.  I think there should be a project to create easy to use administration tools for KDE and/or GNOME that allow the user to configure their Linux system using only the GUI.

Another issue with the desktop is, well, polish.  KDE and GNOME are great, but both (especially KDE) could use an interface polish.  (GNOME could use a speed boost and some bug fixes in Nautilus, but that's a different story.)  It's really not that hard - just eliminate redundency in menus, lay out things in a more simplified and easy to grasp way - in short, don't suffer from feature creep.  And for heaven's sake, include a sane and attractive desktop configuration out of the box.  The fact is that 90% of computer users don't change their PC's settings - ever.

Problem #6 - Multimedia and the DMCA

This is a tough but real issue.  How will Linux users be able to legally enjoy multimedia like Quicktime, WMV, MP3's, DVD's and so on, when practically all of these things are rife with ominous "Intellectual Property" concerns?  There is currently no legal way to watch DVD's in Linux using Free Software, and if current US laws don't change it may remain that way indefinitely.  This issue is really more of a legal problem than a techinal one.  We need to change laws in order for Linux to be usable for the average person in this respect.

Well, that perhaps covers it.  I have been typing for a straight hour now, so I think I'll give it a rest, maybe add some more stuff later.

So what do you all think?

Stryker:
intresting, i wanna read more...

although i've never had a problem viewing dvds unless they were encrypted... and even then ogle took care of it. It is illegal to decrypt them? perhaps what should be done is someone needs to make a good, proprietary, dvd viewer for linux. From what I (think I) know, getting the decryption information legally isn't possible for open source programs because then everyone can see them and it defeats the point of encrypting it. So make a proprietary one. It works for other operating systems why not linux? but maybe i'm not seeing the problem as it really is. I dont think i'm up to speed with all this dvd stuff.

i like ur ideas though, intrested in hearing more.

Laukev7:
I say that the desktop is best left to Mac OS X. FreeBSD and Linux should stay on servers. And maybe Windows for Solitaire and blue screens.

Zombie9920:

quote:Originally posted by Laukev7:
I say that the desktop is best left to Mac OS X. FreeBSD and Linux should stay on servers. And maybe Windows for Solitaire and blue screens.
--- End quote ---



The sad thing is Linux has a larger user base than Mac OS. Macs haven't been able to overthrow x86 in over 17 years(not by a long shot) and I doubt they ever will. Mac OS hasn't been able to overthrow Windows since Windows 2.0 and I doubt it ever will. Microsoft has a good thing going on and I forsee them ruling the desktop world for years to come.

Linux hasn't been able to overthrow Windows since it was spawned(for a long time) and I doubt it ever will. I bet the end of Microsoft will come from an unknown OS from an unexpected company some time from now. The OS that will get the job done will not use the Linux kernel or Unix kernel. It will not be open source, it will not have anything to do with GNU and it will not be under GPL. It will not be rehash from any other OS and of course it will be easy to use. Even though the OS will overthrow MS's dominance it will not knock MS out of the game. MS will just become #2.

Of course I don't know this for a fact but that is how I forsee it happening.

[ November 19, 2003: Message edited by: Viper ]

hm_murdock:
I think Mickey Mouse is going to come out with a good OS soon. It's gonna be called Epcot '04 or something like that. Yeah. That's the ticket.

No, really. All those are valid points... and again, I must point to NeXTStep... it solved all these problems years ago...

Because the solution exists, the problem presents itself. Why doesn't the Linux community try to spur the development of a good X11 alternative? Linux User posted a link to something that looks damn good in another thread. Fresco is very promising. Also, what would have been wrong with someone developing Display Ghostscript?

http://www.gyve.org/dgs/

It apparently exists, but like so many GREAT PROJECTS it dies, and in its stead, mediocrity continues.

As for software installs? Again, need to bow down to NeXTStep...

NeXTStep has two ways... one way is an actual OS mechanism. You have .pkg files which contain an install script, and the files to be installed in a compressed format. What is the .pkg file? Not a file, actually. .pkg bundles, are bundles like any other bundle in NeXTStep... they're actually FOLDERS! Nothing fancy, it's just that the file browser (Workplace Manager in NeXTStep, OpenStep, and Rhapsody, or Finder in Mac OS X) sees them as a single object (remember, object-oriented design!).



You double-click the .pkg and the Installer opens...



It handles everything. There are no dependency issues... EVER. Apps use documented APIs that are provided by the OS. This guarantees that the app will always run, every time, on every machine that runs the OS. NeXTStep doesn't use the low-level UNIX methods for running and installing apps.

Apps are also stored in .app bundles, which, like .pkg bundles contain everything needed. The binary, all of its resources (images, icon, strings, et cetera) are inside there, and like .pkg bundles, it is treated as a single object by the file browser. You see an icon labled "Mail" but it's really a 5MB folder tree.

And that leads to the other way to install apps in NeXTStep... drag-and-drop. If the app contains everything it needs in one icon... you can just move and copy it around. No problemo!

How does it do it? NeXTStep is really, really layered. There's the low-level stuff, and then atop all that rides the high-level Objective-C stuff. All the high-level components interact with the kernel directly, and not as applications. YellowBox and Cocoa would be hard to run on Linux without an AWFUL LOT of work being done. See GNUStep... they're building NeXTStep-compliant components to create a NeXT-layer for other OSes.

They've done an admirable job, btw.

Navigation

[0] Message Index

[#] Next page