Fog Creek Software
Discussion Board




Is Linux as really serious threat on the dekstop?

Will it ever actually happen, that even small companies i.e. less than 10 staff will choose to install it on their networks instead of windows or do you think that windows is too ingrained?

Thoughts?

Mike
Wednesday, January 01, 2003

Having written the below, I realize I speak about Linux as if it's more than just a kernal: note that I am referring to Linux as the kernal, utilities, graphical system, desktop environment, etc.

That's a hard question. I'd like to think so.

My reasoning is simple: Five years ago, there was no KDE. Well there was, but it wasn't what it is today. Not anywhere near what it is today.

Five years ago, there was kernal 2.0.x. That series had poor SMP support (not improved until 2.4). No USB support (was that added in 2.2 or 2.4? Hmm...). There was no GNOME.

My point is that they've come farther in their time than anything else I know of. Compare it to Windows: Windows took a lot longer to come as far as the currently availible Linux software.

It's not perfect, not by a long shot, and it is filled with its own deficiencies (Like everything else, but perhaps a few more). But it's getting there. And when Linux gets something right, it does it really well (Of course, Unix printing, say, has always been atrociously horrid. But multi-head works amazingly well).

And as for software, I think there will be more software coming. My reasoning is simple: Much much more now has been ported to Linux than before. This is still a very small number, but much bigger than before ("If we take a penguin, and enlarge it to be the size of a human, we see that its brain is still smaller than the human's, but - and this is the point! - it's bigger than it was!").

In addition, MacOS X being a Unix now makes it easier for companies to port things to Linux. Sure, with Cocoa and such there's still GUI toolkits that need to be abstracted and work to be done, but going from OSX -> Linux is a smaller gap than Win98 -> Linux.

So basically, I think it CAN be a serious threat. Do I think it WILL be? I think it's garnered enough mindshare that eventually it will be a contender. I seriously doubt it'll ever take over, at least not for a *long* time (look at Windows 95! Still out there and running strong for many people!).

If it continues to improve at the pace it has set, it stands a real chance out there.

Mike Swieton
Wednesday, January 01, 2003

Regarding *small* companies and individuals using software on the *desktop*, no.

For small companies, they upgrade their hardware as often as they do their OS (every 3 years or so) and windows comes 'free'. Sure, they pay an OEM fee which is insignifigant. Linux advantages don't kick in economically untill you have many more computers than that. Also, for a small business there is a larger supply of Windows experts to set it up than Linux experts so doing installation is the same cost. Linux costs more to begin with since in addition to the OEM Windows, they pay for the Linux CD at $85 or so, which is more than they paid for their OEM Windows. And now they can't run MS Office which is the basic desktop application every business desk needs. Pretty soon you have to ask what's the point.

For home users other than Unix fans and hobbiests, why on earth would they choose Linux? It doesn't make sense for most people. Windows comes on their computer with Office and it works when they plug it in and after a few years when it's not working because of all the device conflicts, they buy a new computer so they can play "Killathon 2006" while wearing their special four-dimensional viewing helmet.

Linux can compete as an inexpensive replacement for proprietary Unix on web servers and computation clusters. In the web server market, it's competing with OSX. Those two are having last rites for the venerable Solaris. Linux is a great way for Unix fans to have inexpensive access to their favorite OS. It might penetrate at very large organizations that can afford a big computing staff but who don't like the cost or restrictions of the licenses. But on the home desktop or at small businesses I don't see it happening.

X. J. Scott
Wednesday, January 01, 2003

Mike, are you trying to start a forest fire.  Other topics to choose from include:
Best text editor
Best Browser
Code commenting
Best processor architechture
Windows.  Any good?

Smokey da Bears
Wednesday, January 01, 2003

Ha ha. You do realise that you have managed to post three potentially explosive questions tonight. :D This one would be considred flamebait on most forums, for a very good reason.
Nobody knows. Some claim to know, but its wishfull thinking. Linux is a very unpredicable entity.
I believe that in order for Linux to be a serious competitor a few things need to happen.

MS must generate a bit more discontent in their customers.

A big, powerfull company must release a linux desktop distro with a killer app. (Think Adobe Linux + Photoshop etc)

This distro must have a desktop environment that is more consistent and less complicated than the current big ones.

I dont think it will happen though. My hope stands with OpenBeos.

Now, I bet someone will disagree with me and we will get in to an angry disscussion over what everyone knows is pure speculation. :D

Eric DeBois
Wednesday, January 01, 2003

Sir, I don't agree with you one bit.  I think Linux kernel 3.4 is gonna stomp OpenBeOS.  You are just wrong.

Something like that ;)

Smokey da Bears
Wednesday, January 01, 2003

"MacOS X being a Unix now makes it easier for companies to port things to Linux. Sure, with Cocoa and such there's still GUI toolkits that need to be abstracted and work to be done,"

Is work being done on this? Is it possible legally to copy the Cocoa API?

"but going from OSX -> Linux is a smaller gap than Win98 -> Linux."

I have seen several apps go from Linux -> MacOS successfully but haven't seen the reverse yet. Is it happening?

X. J. Scott
Wednesday, January 01, 2003

>Best text editor

emacs of course

>Best Browser

unanswerable since choosing one means choosing the least repellant evil

> Code commenting

yes please, but keep it synchronized

>Best processor architechture

The 6805

X. J. Scott
Wednesday, January 01, 2003

On the desktop, Windows and Linux basically cost the same amount: zero.  What's $100 for Windows and $200 for Office Lite relative to an employee's salary?  What's it worth to be able to send someone to Word/Excel training?  Having them use the same GUI than they use at home?  Having to scour the planet for a Linux app to do some quick task?

I can think of a handful of reasons where Linux isn't $300 better than Windows, but only a handful.

Another consideration:  Microsoft is rather fond their dominant position and if cost becomes the driving reason people are switching to Linux, they'll just reduce the cost and take it out of their $50B cash hoard.  I can't see them just letting their market share "wither away" without attacking the fundamental reasons people are switching.

If times get tough, look for MS to open source Windows, cut the pricing, offer it ala-carte, improve stability, etc.  Likely?  No, but things haven't gotten that tough yet.

Bill Carlson
Wednesday, January 01, 2003

If you want "Linux will eventually rule the desktop" arguments, go to Slashdot, where the rhetoric was invented.

Having used Linux extensively on the desktop, all I can say is this:  if the current Linux/Desktop user experience is satifisfying, you are easily satisfied.  It is like Ham Radio - fun to tinker with, but if I want to just talk to someone I'll just call them on the phone.  Linux/X11/KDE is fun to tweak, but if I want to get something done I use Win2K.

'Cause it just works, and works well  Right out of the box.  Time is money, after all.

Mitch & Murray (from downtown)
Wednesday, January 01, 2003

I remember reading headlines a couple years ago saying, "Linux on the desktop is dead."  What a ridiculous thing to say, then or now.  Linux on the desktop will be dead only when all the major efforts to build Linux desktops and office applications are abandoned.  It doesn't seem likely any are going to be abandoned in the near future.

As someone else remarked, the main open source projects devoted to linux desktops and office apps are not very old.  KDE and Gnome are both less than 6 years old.  They've come quite a ways in the last few years, and they're still plugging way.  No, they're not there yet.  But if they continue making the same progress they've made in the last few years, it seems likely that they're going to be come a threat to Windows on the desktop sooner or later.

It also seems obvious that whenever the Linux desktop does get there, it will offer a compelling price advantage over Windows.  Just look at the $199 Walmart PC's with Lindows or Lycoris Linux installs right now.  These desktops aren't serious threats to Windows.  But there's no way anyone can sell a $199 pc with Windows either.  When the Linux desktops do improve, they'll have the same price advantage.

Herbert Sitz
Thursday, January 02, 2003

I think Linux works well mainly as a server. It is a reliable operating system, and people boast about how long they can leave a Linux box on without any need of a reboot.

But I've always found Windows has a better user interface. This is possibly because a good set of program managers sat down to work out the best possible UI spec, which isn't really feasible on an open source team. I've never really felt as "in control" using a Linux GUI, which is probably why I still control things mainly straight from the shell.

Better than being unemployed....
Thursday, January 02, 2003

Sorry if these topics may be considered as flamebait.

That is not my intention. I am at a crossroads and I am trying to decide whether cross platform is worth it or if is better to stick with being windows only. Thats all.

I have been stuck in a niche market for about 6 years and I am trying to climb out and choose something mainstream.

Mike
Thursday, January 02, 2003

" I am at a crossroads and I am trying to decide whether cross platform is worth it or if is better to stick with being windows only. Thats all."

What would you consider "worth it"?

Make more money? Most often no.
Be cool on /.? Most certainly.

... or everything in between.

Just me (Sir to you)
Thursday, January 02, 2003

Linux on the desktop will not have a supportability advantage over Windoze in the end.  It will not attack the larger problem with desktop OSes of deep dark corners that only experts dare to tread that happen to cause great problems.

That being said, it takes somebody who is motivated enough to shove it down people's throats for it to become a threat on the desktop.  Because most people do what they are told, with respect to computers.

I have a very large nervous feeling about BeOS/OpenBeOS/AtheOS/etc.  They are making very nice object oriented APIs without layers of indirection.  What makes me nervous is that they are relying on the vtable layout of objects in memory, which doesn't have the assurances of COM and CORBA.  I just have a bad feeling that they're going to run into major problems down the road, if any of them take off.

w.h.
Thursday, January 02, 2003

I'm wondering of Apple's Darwin will make any in-roads on the current Linux market. It doesn't seem as yet that any non-Apple groups have made much of an impact on Darwin.

pb
Thursday, January 02, 2003

I'd wager that a rather large percentage of those $199 Wal-mart Linux boxes end up running Windows, even if it's a pirated copy. If a computer can't run MS Office, IE, and $COOL_GAME_OF_THE_MONTH, it has very little appeal.

Dave Rothgery
Thursday, January 02, 2003

The main reason we are likely to see more Linux on the desktop is that MS can no longer legally prevent OEM's from partitioning the HD and installling a dual boot configuration. Previously an OEM could not even install Windows and put the  docs and user settings on a D partition so that the reinstall disk wouldn't wipe out all the data.

Secondly the cost of computer hardware has dropped whilst the cost of MS software has remained static. The result is that even an OEM copy of XP Pro,  MS Office and a graphics suite cost more than the machine and monitor combined. We are likely to see Linux installed on these machines, and of course on set top boxes and other dedicated machines, in order to pull in  those at the bottom of the economic scale who haven't yet got a computer, but wouldn't mind accessing the internet occasionally.

People who complain about Linux being difficult to install are forgetting that their grandmother would be incapable of installing windows as well. In fact Linux is now easier to install than Windows, since you can install everything at one go. The problem has always been drivers but with a manufacturer's decidacted Linux machine  that will not be a problem.

Somebody brought up the question of piracy. They have latched on to a little dissemintated truth. MS loves piracy because it is the best free direct marketing tool it could possibly have; it doesn't have to pay for the materials or the distribution, and it can change it's policy and ask for all its previous "partners" to be jaiiled at a moment's notice.

The fact that Linux has taken off in Jordan as a result of the BSA putting pressure on the government is proof that piracy of MS products is the best weapon against Linux MS has. (In Saudi the only software you can't get for $3 a CD is Suse Linux which you have to buy in the boxed set!).

Stephen Jones
Friday, January 03, 2003

If the real question is Cross platform versus Windows only I would say go for Cross platform for one reason:  I will force you to design better.  You will want to get as much of your code shared between two systems as possible, with the only things different being those that are truely different on two platforms.  See the Bridge Pattern in the Design Patterns book for a reference.

Think about how you would design it for the following three scenarios:

1) Desktop application
2) Web Based Application
3) Command line application

Before anyscoffs at 3, think about a minimum configuration script designed to be run over a dial up connection via a data center back door, or as part of a larger scripted install process.

IF 2 and Three solve your problems, then you probably don't need to pursue a cross platform UI toolkit.  If you go web based, and you can pick a cross platform server you'll have better flexibility as well.  This is where Java excells, but You may decide that you need something that integrates tigheter with C code, so you could do PHP, or write a custom Apache module.

If you need a richer application environment than the web provides, and you want Cross platform, and you need the performance of a C/C++ setup, you will want to find a cross platform UI toolkit.  Qt from troll tech is what KDE uses , so you know you will cover a good portion of the Linux/Unix space.  It runs as a DLL on windows, so you will cover your primary base as well.

Adam Young
Saturday, January 04, 2003

As for a small company installing it as there primary, there is no reason why not.  I've done it before, albeit for a web development shop where the server platform was linux.  But a good portion of the company still used Windows for document preparation, and Foxpro development.

The amount of software available for free (www.freshmeat.net, www.sourceforge.net) is staggering.  Most of it is targetted at Linux first, and Windows second.

One nice thing you get from Linux for the small office is X Sharing.  It makes it trivial to have a desktop app installed on a single machine that can be run by everyone.With Windows, you need terminal server, PCAnywhere, or something like that to share a windows app.  Or ,it the app is a X ap, you need a Unix/Linux Box.  Or perhaps Cygwin although I have not tested that.  X Sharing can be abused. It still has licensing issues around number of concurrent users.  It requires a decent network setup since all UI goes across the network. 


Another thing Linux (especially Debian) has going for it is ease of administration.  Yeah, the setup can be a bitch, but the ability to then update all packages for security is quite easy.  I don't have much Win2k or WinXP experience, so I can't say if they are comparable, easier, harder or what.  BTW, the reason I said especially Debian is that, even though more stuff is set up for RedHat's RPM, the debian package management setup is the best supported in Linux land.

So I would sa y a small technology company would have no problem going Linux only.  A small construction company, that uses Excel macros and custom Windows based software estimating tools would have a hard time.  But if you make your tool crossplatform, you can sell to both.

Adam Young
Saturday, January 04, 2003

And if you create a command-line interface to your app, you can easily automate testing.

runtime
Monday, January 06, 2003

*  Recent Topics

*  Fog Creek Home