Ubuntu man challenges open source to out-pretty Apple

one can't 'out-pretty' Apple

If the standard of 'pretty' is what Apple does, then -as time has shown-

only Apple can be what is pretty. All Job-sonian true believers will never

accept anything else.

The End.

development tools

Make Eclipse better than Visual Studio 200* editions (more functional, faster, providing integration with OSS AND proprietary source control systems), including better debugger for C and C++ code. Without it few ISV will consider writing for Linux.

I don't see it...

Linux is produced by a certain type of person for a certain type of person. It does its job - much the way a John Deere tractor or Mack truck does its job. That's fine.

The Apple experience is created by 'design' for people who don't care about computers and never open the bonnet of their car except to fill up the windscreen washer fluid. They expect it to 'just work'! They get no joy out of tinkering and value ease-of-use and reliability more than speed, horsepower, adaptability - and indeed price.

Horses for courses.

The key word here is 'design'. Design has two main aspects. 'Functionalty' and 'Styling'. Functionality is about doing what it is supposed to. Styling is the aesthetic presentation. You can have one without the other but it is less good than getting both right.

Linux scores pretty highly on functionality - provided you know what you are doing and are prepared to work at it. The very fact that 'open source' is being challenged to 'out-pretty' Apple completely misses the point. It's not about prettyness, its about design. Prettyness is superficial styling. It is eye candy. It does not improve the user experience and misused, as is often the case, can actually get in the way and reduce functionality.

I am more than happy to use Linux on my Web server. Ubuntu on the desktop I find butt ugly both asthetically and functionally. Why try to be a cheap imitation of Windows or a prettier version of Mac OS X?

Do what you do best better and the World will flock to your door.

I maintain...

That one of nux' deadliest shortcomings is that nothing yo buy in a box comes woith software for it. People spend fortunes on camera's, mp3 players, phones, organizers, blackberries, printer multifuctionals anf GPS'. Pretty soon you'll be finding a driver disk with your fridge !

And NEVER is any 'nux stuff included. End EVERY time Linux supporters jump up and down that stuff can easily be found on the net, and there's wine etc.

People want a CD in the box with the stuff they just bought, one you chuck into your pc and it installs.

Als long as the 'nuxers refuse to see an OS as a means instead of an end it'll never happen. Pretty or not.

Peter R.

@They Don't Get it

Like any old Mac user would know, there's a difference between a disk being unmounted and ejecting the disk.

In those days of floppy disks, you could eject a disk, but still leave it mounted - the OS would ask for the disk back as needed. To actually unmount the disk, you needed to drag it to the Trashcan.

Standards

The problem with Open Source is that dispirate coders come up with different solutions to common problems (and then usually call said solutions something stupid. YaST comes to mind).

What's needed is a community based *solely* on design before coding ever takes place; a community that doesn't worry about the code itself, but specifications of the product and it's interoperability. A community to create an international standard for operating systems, if you will.

Now THERE'S a pipe dream...

OK..So

Here's a UI improvement suggestion for you: How about making it so I don't have to use the command line to join a wireless network? (Xubunu, downloaded last week, netgear wireless card based on a realtek chipset)

Seen it all before

How about going back in time maybe 10-15 years. The widget wars were over and petty much everything was based on Motif. A little conservative, perhaps, but fast, consistent, and it worked. CDE was becoming much more mainstream giving the user a desktop and further standardising the interface.

Compare the situation now - you still have Motif for some apps, others are KDE, GNOME or something else. GNOME and KDE are huge codebases in their own right and since you typically need both that is the root cause of much of the bloat present in modern Unices.

The answer then, is not to introduce yet another UI standard. It is to settle on _one_ of the existing ones. Personally I don't give a fuck about eye candy, I want apps that work well and work consistently first and foremost. This is something to bear in mind whenever someone advocates a new UI metaphor. It doesn't help that they invariably waste twice as much screen space as needed and look like they've been created by a five year old.

@Matthew

"Already prettier than apple and based on Ubuntu."

Hmm ... you're not familiar with Apple's Front Row then? The Eearos developers certainly are.

pretty desktops?

KDE comes with pretty as standard and alot of other things come with more pretty.

What Linux lacks is an out of the box double click experiance.

What do I mean? Well in windows at least you double click a document and it opens in something (or tells you that it can't as it doesn't have the stuff needed)you double click an mp3 and it plays, you stick in a dvd and it plays, you double click an avi and it plays. You double click an install file, it asks a question and it installs. You stick a disk in the drive it auto runs. You download a file, double click and it runs.

Simply put any monkey can get somewhere with a windows machine with little thought or effort - the world is just a desktop icon and a double click away.

Linux, is not, and until it is it shall never be widly accepted on standard desktops and laptops.

Some rules for open source to follow

Thinking more on the subject, I've come up with some rules (in no particular order) that Open Sourcers might want to adopt:

1. Command line operation should always be an option, but never a requirement (a GUI gives the user a chance to figure it out for themselves. With CLI you either know it or you don't)

2. Interface working methods should be as uniform as possible, regardless of objective

3. Always assume the user knows NOTHING

4. Functionality is not enough; easy access to functionality is everything

5. Document everything, and get it proof read!

There are many more, but there's five to start with. They would increase the overheads of open source development significantly, but that's the price to be paid for an OS that can truly compete.

Leave the pretties to the artists. Not developers.

@Drak

/usr/local or /opt

Don't quite get your point, NWN & UT2004 work perfectly across any distro. They come as a complete package and expect to find nothing more than a value stuffed into $HOME.

However, you do end up losing out on the advantage of common libraries like this. But, it's easily fixed.....

Each distro should hold a file in /etc called something like 'locations' (or whatever) and list where an installer is expected to install the appropriate files.

Likewise, have a common db that local installs can list what files they use, so the package manager can check it's own db (regardless of format) and the 'local install' db to see what files are loose. It would be simple to write a script to handle the install of ANY software then.

I admit some distros do funny things with certain config file locations, but most commercial software shouldn't be touching these config files any way (but, can also be resolved by above method).

@Gerhardt

Never seen an update for UT2004 for upgraded Libc, and it still works (he says after not playing it for some time, but meh - never seen an issue like that). So I'm guessing it shouldn't be a huge problem. I am willing to stand corrected here as I am nowhere near being an expert on libc.

Kernel modules suffer from it though. - So, thinking about it, provided the libraries are 'static' to the program and aren't anything to do with the system you really should be fine (MS _probably/may_ still use the same compiler+faults from years ago).

How about MS (that's Mark, not Micro) puts some effort into making pretty frontends to stuff like firewalls etc that work on X, and NOT KDE/GNOME. I like the layout of UDE (I would recommend to people who like *different*), therefore don't want all the KDE/GNOME base stuff installed. Autologin for xdm? available for kdm and gdm, but could do with being on all home systems I feel.

It's the little things that need fixing

Like the fonts, the geeky language, the names...

"The Gimp" - who the hell would use that over Photoshop? I'm not talking about functionality, I'm talking about not getting funny looks when you mention it to people.

Gnome? Awful name.

And the different Ubuntus (Xubuntu, Kubuntu, etc.) just sound dreadful.

To geeks, techy stuff wins out every time, but I'm sure a big part of Windows success is it's name. Names like Xubuntu or Gnome or Puppy Linux just don't sound good.

Linux drivers with gadgets

Clearly Peter R is unfamiliar with the massive range of "works out of the box on Linux" devices (cameras, mp3 players - practically everything worth using, in fact).

Even so, a large proportion of IT items I buy these days come with Linux drivers on the CD, or give instructions on installation under Linux.

Most Windows users are addicted to shoving CDs full of badly written drivers and awful, bloated and downright unnecessary apps into their machines (I know this, because I often have to sort out the damage they cause).

@Peter. R. : too late ...

If you buy an LG fridge with built in TV and web surfing capabilties it does come with a driver cd ...

The trouble with Linux...

... (or at least one of the troubles) is all of the petty infighting:

"My desktop is better than yours" KDE vs. Gnome vs. Enlightenment vs. a badger's arse;

"My installer is better than yours" rpm vs. tar.gz vs. god-knows-what

Just think. what if all of the Linux developers (and I mean that in the loosest possible terms) actually had a common roadmap? Actually developed to some core standards (directory structure, device naming, 'window manager', fonts, etc. etc.)?

I can't see that *ever* happening. And that's why Linux will never get to be the mainstream desktop platform.

so why doesnt Shuttleworth buy Borland IDE

Everyone knows that the Gnome/KDE GUI toolkits are a mess, so why doesnt Shuttleworth put his money where his mouth is and create a common GUI standard for Linux. Borland has been trying to sell their IDE product division for a long time, that could be a pretty high end starting point. Its already proven as fast and streamlined. And maybe it could be integrated with Eclipse.

a Reg article about Borland can be seen here:

http://www.theregister.co.uk/2008/07/10/david_intersimone_embarcadero/

@Edward Rose

----"Likewise, have a common db that local installs can list what files they use, so the package manager can check it's own db (regardless of format) and the 'local install' db to see what files are loose. It would be simple to write a script to handle the install of ANY software then."------

OK, sounds good, so why doesnt someone in the Linux community do it? I say someone (like Shuttleworth) and not everyone because the average open source dev is only interested do-it-yourself geeky devtools, and not serving the general public. Shuttleworth is going to have to shoulder the burden himself if he wants anything serious to get done.

@It's the little things that need fixing

Joe hit the nail right on the head. It's time to drop the stupid naming conventions and the forced acronyms. To me Gimp is a character from Pulp Fiction not a graphics package. Same to 99% of the rest of the planet (well those who've seen Pulp Fiction). Likewise I will not use a Word Processor called "AssRape" or a Spreadsheet called "SheepShagger". And any company that uses the following naming convention: [colour][animal] can fuck right off too.

The Linux GUI designers problem...

... is that they don't understand the point of them. The main objective of a GUI is discoverability. A secondary objective is portability of knowledge.

Simply having a graphical box that allows you to enter the same text you'd use at the command prompt does not count. Similarly, putting all the options that a program supports in a drop-down list isn't good enough if it still means having to read a man page to figure out what is going on. And especially not if most of those options aren't applicable to that particular user on that particular machine.

All technical language should be avoided as far as possible. If two programs do the same thing then they should do it with the same interface - consistency is king. Subtle communications (like the way the red 'close' dot in OS X had a darkened centre if that window has unsaved work in it) are better than conspicuous ones if they relate to things that the user may or may not care about. Conspicuous communications should only be used if you need to communicate something critical.

Modal interfaces are generally to be avoided. Use context, but not as part of the main program interface. So inspector palettes are good, ribbon interfaces are bad.

Never, ever use the multiple document interface. People prefer to think in terms of tasks and documents. Forcing them to hunt down their task by program first and document second is not helpful, and in any case it unnecessarily complicates the desktop metaphor and the normal ways of locating what you're doing.

Feedback should be offered on everything. Even Windows fails here - the default behaviour for keyboard shortcuts like ctrl+c, ctrl+v and ctrl+s is usually to give no feedback. Office adds a collapsing box for ctrl+s but that's an application-specific hack.

If it's likely that you know what a user wants to do then if it isn't otherwise jarring, just do it. But don't go too far. And don't add f*cking talking paperclips to your office software.

if you want to pay for software

feel free.

just don't tell me that because you've paid for it it automatically means it's better (trundle along to your favourite encyclopaedia and look for 'cognitive dissonance').

As for Vista running programs that are 8 years old - wow, must be good if it runs ALL 8 year old programs.

No doubt it operates all those 8-year-old periprerals as well?

@@@Matthew "The solution is here" By J

No, used it today (and everyday for the last 6 years) and despite being pretty it's still crap at being a desktop system.

Double-click installation

I've got a *IX that has a nice package manager, a standard installation format, and even files you can double-click on from the OS to install them. It even had a nice-looking GUI.

Think I'm talking about OS X?

Nope.

Silicon Graphics had most of it right back in the 90s with IRIX. Sure, it was expensive, and it only ran on their hardware, some of the time, but it was far more polished than Linux is, after a further 8-10 years of development. Download and double-click a .tardist file, and the Software Installation program would launch, run through the prerequisites, and install the software for you. (God help you if any prereqs were missing though...) It worked for both open-source and commercial software, and even handled OS upgrades and patches.

The free-tards need to get an old Octane and take a long look at it.

Mine's the one with the pipe-cube on the back of it.

Re: if you want to pay for software

What is wrong with paying for software? Software takes time and skill to write. Time costs money. People have to eat. If you work for nothing, give me your contact details - I have a lot of jobs that need to be done! Nah, forget that, I'd rather have them done properly than for free.

@Thomas Davie

Unmount is no worse than Eject. Eject makes sense for a CD or DVD, but makes little sense for a USB or Firewire drive. The only advantage is that it can have a nice recognisable icon. It's harder to make one for 'Release from the clutches of the OS in a controlled manner' or "Liberate Resource to allow safe disconnection"

It needs to be simple and consistent

I've just finished a project where the client asked me to set up a Java dev lab, so I used Linux on the servers and Mac development clients, which was a thoroughly pleasant experience. No BSODs, everything in it's right place and a place for everything.

I started a new project on Monday and the client had provided brand new RHL clients. I dislike Windoze intensely, so I desperately wanted this to work, but even as a relatively competent sysadmin it took me a day and a half to finally get to a KDE desktop that didn't look like a toddler had designed it, with all the right packages installed. Then I fire up Firefox and Eclipse and both apps totally ignored the KDE settings and were in "butt-ugly mode". Yes, I know there's a GTK fix, but this is the sort of thing that makes average users run a mile.

As for the other guy who started the same day as me, he's never been near a UNIX box in his life. Understandably, installing Tomcat, Apache, the Ganymede release of Eclipse is a real challenge for him - every time I winced when I felt like saying "just drop into a command prompt and su to root, configure, make and install."

As others have said, when the GUI can install any app in a simple manner, prompting for the admin password at appropriate times, and everything looks and works the same then we'll be half way towards a Mac OS X experience.

@Colin fcukin Morris

Technically, hardware tweakery is not really the fault of Linux. I have a laptop running Ubuntu that does not display the problems you mention (using it right now). If your laptop vendor made the appropriate Linux drivers etc, then these problems would go away.

But, from a practical perspective, it does not matter who is at fault. Ultimately you cannot make your laptop work with Ubuntu. From a Joe Sixpack perspective, the laptop works with Windows, but not with Linux. Therefore Linux is broken.

The real problem is disconnect between how techies view the problem and how users see the problem. Also, those that could fix the problem are not very motivated to do so.

Now hardware vendors typically have a hard time keeping their heads above water supporting drivers etc just for Windows. Why would they want to more than double their work just to support Linux - with 1% or so marketshare? That just does not make business sense for many.

I expect it will be a few years yet before the end user experience with Linux is as slick as it is for Windows.

All mouth and no trousers.

I remember there was a mass appeal for designers about 6 months before Hardy was due. Being not so bad at it myself (but not fantastic) I came up with a few designs, refined them based on rather limited user feedback*, and submitted them to the art mailinglist + wiki, and...

... nothing happened. A few people said 'that's nice' but thats about it. Nobody with any control over any form of art in the distro had anything to do with the art mailinglist. After some pushing it was made clear that the only way someone whould ever look at it was if I coded it into a theme myself and even then the best I could hope for was some comments on gnome-look.org. eff that.

If Ubuntu was a customer for me in my day job I would have told them to piss off by now, and that is if they were paying me in the first place. I have no idea how the hell they intend to get anything decent out of any artist if...

1) They expect them to create the theme, not just the graphics

2) They expect them to do the above with absolutely no feedback along the way and

3) They don't even have anyone in a position to do anything about it if they do get something decent.

With a normal client I'd get feedback based on the work so far, I certainly would never deliver a project without client sign-off at multiple stages, yet the Ubuntu art project expects designers to give them a fully functional theme package before they will even bother to say what they want.

I for one am certainly not wasting days wrestling with bizarreo theme generators only to find out 'nope, we don't like it'.

Which brings up the final point, you cant gain OSX style usability with just a skin job, it needs to be started a base level, graphics are just the trimmings, and if they obviously dont care about even the graphics then, I am sorry, it's always going to suck.

* Feedback was gained by spamming non-related forums as the official sources were useless.

One part is a Linux fault

Every so often, the interface for Linux drivers is changed. This is deliberately done to break old drivers that are distributed only in object code form, and it is done at Linus Torvalds' insistence.

I know some people would like a system that is made of pure open source. But most people out there would like to be able to buy any make of video card on the market, and have it come with a working driver for the operating system they would like to use - even if it's Linux, and even if the manufacturer wants to keep some of the card's goodies secret.

So this is Linux shooting itself in the foot, IMO.

And if third-party developers can't distribute binaries that will install on any Linux computer that meets basic conditions - an i386 CPU instead of Itanium or PowerPC, Linux version later than whatever point zero, has to include a copy of KDE installed - then there's not much chance of going down to your local chemists' shop (big discount one that also sells computers, of course) and seeing lined up ten different versions of the latest game for ten different distros, is there?

For ordinary users, who aren't going to write their own C or Fortran programs, buying a Linux machine is almost like going with WebTV instead of a computer. Linspire tried to do something about that, but I'm not surprised that the way they chose to go about it hasn't yet worked out.

Linux is wonderful, say, for people who want to set up their own servers, but for the mass-market home desktop, it doesn't meet the need, even though Linux advocates are right that it comes a lot closer to doing so than people perceive it to. Trouble is, close isn't enough. So the Mac, rather than Linux, will benefit if there are serious enough problems with Vista to shake Microsoft's dominance.

Lack of Accountability

As people have discussed here, there are many, many issues involved in making Linux a viable consumer OS. The largest obstacle to solving these problems is simply this: There is no single person or team of persons ultimately responsible for coordinating and integrating the solutions, and polishing Linux into a consumer-grade operating system platform. Open source is a ragtag collection of volunteer developers and geek heads, and together they've accomplished all sorts of technical wonders. It's also, by design, a competitive ecosystem that leads to hundreds of distros and software branches. But it is the very democratic (or pseudo-democratic) nature of the open source community that prevents Linux from making the jump from a utilitarian, engineering tool to a commercial, consumer-level product. Consumers expect, and demand, that their computers be as easy to use and problem-free as a toaster, television, or automobile. Open source doesn't have the genes to make this happen. There must be a controlling decision maker or leader, and there probably will never be one, because of the inevitable infighting that will ensue. Democracy sucks.

sounds old ?

sounds like I've heard this a while ago.

@ Tim Greve

Install likewise-open-* for AD.

I'm terribly sorry...

I completely misunderstood the problem. It's not the things I mentioned before ! Well, not all of them anyway...

It's the zealots that keep ignoring users and maintaining that everything works right away in Linux, and that all software is easily available, and on and on.

Listen up, wise men of the Penguin ! I tried a new install of buntu a few weeks ago. I even dedicated a machine to it. And then I tried to plug in all the silly stuff I own into it. Oh, and the first Linux I installed was Suse 6. I still have the box.So I'm not a complete newbie.

And things are NOT as keep shouting. After a lot of work I did get a few things working...for a while. But 'normal' users would have thrown the distro in the bin long before I switched off. Oh, and I do the experiment every time new major releases appear. And always I hope for the best. And everytime I'm disappointed. Not because I can't get it going. But because I know that, again, the bulk of the general public wil be reverting to windows. After about 15 minutes.

But hey, never mind me. It's just me. Please ignore me, it's not important. But if you are serious about wanting to make Linux a serious alternative for the non-geek you beter start sitting up and taking notes.

Peter R.

Thats the trouble with consumers

- they want it to look pretty. That way they dont have to think about functionality. The IT industry has been selling style without content for too long.

As an ageing hacker it amazes me that most companies now have thousands of PC's that are millions of times faster that the first computers but very few can use them as effectively to help their businesses than Leo.

We've come a long way since then. Its prettier here - but a lot less functional.

GNU+Linux it's here it's now and it's usable

First a disclaimer: I'm a developer of a GNU+Linux Distribution(and no it's non of the common ones) but I have to deal with normal(those who are used to windows) users of GNU+Linux

often(family,friends)

Well as Linux is ONLY the kernel of the entire OS it really shouldn't have any pretty interfaces etc...

Now a GNU+Linux system that one actually has things to use as a desktop.

I tried a few things for my mom(50+, can actually use a computer) over the years for her use.

It ranged from:

GNOME, KDE, XFCE4, Enlightenment to more simple solutions IceWM, Fluxbox etc...

I gave her such setups and let her use them. There was always something missing. I.e. not enough

glue to make it all seamless etc...

So I recently tried the same using ubuntu/kubuntu/xubuntu..

Suffice it to say she is now a very happy user of xubuntu and is more than capable of doing

99% of the operations herself(hell she even updates the system herself).

I asked her every so often if she has any problems or such. And most often the response is

no and at times(which actually suprises me) is this actually works better than on Windows.

There might be a I would like an app that does this or that question so I install that for her

since I tend to know more apps then her.

So I belive GNU+Linux on the desktop is there. What needs to be done is only to improve on it.

I seriously like the way it asks when a certain codec or such is needed to play a file on how to install it.

To this whole notion of common releases etc...

It would make projects stagnate since they would only release like once a year in a staggered release fashion and bugs would increase. The only way to reduce the bug count is as ESR(and no I don't like the guy much) says Release early, release often...

To the make vendors provide GNU+Linux binaries for their apps(i.e. Adobe&co)...

They are free to do so.. infact they could just work with WINE to provide nice easy wrappers around their tools(that is what wino basically is, it's a windows toolkit) or have a common core then just provide Free Software wrappers around that core to compile a launcher for it.(similar to what the nvidia guys do with the drivers they have a common core and wrappers around it to fit into the kernel). This would solve most incompatibilities almost at once.

To the we want a windows type installer thing...

Well there are a few projects in that regard but why bother... Most binary distros nowdays have it setup so that if you double-click the .rpm/.deb/.whatever it will install it automagicaly. Why bother with paths the filesystem is hiearchical... if you need more space distros should provide tools to migrate a certain hiearchy from one disk to the other and then mount it for you automagicaly.

To the should follow common standards for various things:

a) directory locations

it's called FHS - http://www.pathname.com/fhs/

it specifies most directories on a GNU+Linux system and explanations what they should

be used for

b) device naming

it's been standard for a long time... only with udev people have started to botch with it

all my devices are always the same on all machines and that's the standard names they have

c) window managers fents etc... etc...

There is http://www.freedesktop.org - they are working on common standards for the desktop

and have for quite a while. This includes fonts and even RAW file handling from digital cameras.

d) backward compatibility

actually quite a few projects try to work that out but there are inevitable changes with it

even microsoft has that... Just think of all the msvb4,5,6 etc... runtimes one needed.

But as on windows it's the same on GNU+Linux just install the older library along the new one

and you're done.

So as you can see there are things that are being worked on or specifically that have existed for years. I still use what I use out of my own personal choice(that would be fvwm, a handfull of terminal apps and a graphical browser that actually behaves consistently(kazehakase) unlike other browsers that get changed by JS/plugins/etc...).

I'm fed up with all the crap people post... If you are incapable of using a computer don't use it.

If you want Joe Average the technophobe to use a computer give him something like a minimal 3 app computer like the EeePC.

General computing for the average user is dead... They need specialised computers that do only what they want with the ability to upgrade when needed... But of course this isn't in the interest of vendors that then can't flog pricey kit onto computer iliterates then sell them classes, support, software etc...

Ok can someone explain...

serious question....

If I buy some kit, that a friend has, that works happily on Windows XP home edition, I'm pretty sure it will work on XP pro & Server 2003.

So if said firend has same hardware working instantly on Unbuntu, will it work on another version of Linux, say Red Hat or another distro (what ever the bloody hell one of those is)?

If not, why not?

Simple question.

@ Stu Reeves

In Linux, all application programs are by design isolated from hardware. Only the kernel can talk directly to hardware; application programs see a unified interface representing idealised versions of the hardware. Actual hardware drivers must be compiled as loadable kernel modules or part of the kernel.

If your friend has a piece of hardware running on Ubuntu, it **should** run on any other distribution; but it may require a kernel module to be compiled, or -- **in the absolute worst case** -- the whole kernel to be recompiled. If your distribution is reasonably up-to-date, though, the chances are that it will just work.

It's still a long way from matching Windows/OS X user experience

Heck, I echo what an earlier poster said : look at Irix. Yes, now it's rather creaky, but it's well integrated and shows the sort of base experience that can be built upon.

However, as a large proportion of Linux distributions won't even load in VirtualPC without hacking X's colour depth, I think there's some way to go, even if ubuntu is quite pretty. I recently went back to the first Linux distribution I used in '92 - Slackware, as it generally seems to favour functionality over cleverness (like insisting on a graphical install when really it isn't necessary).

Still, I don't think they've got a hope in hell. Most commercial developers don't integrate their apps properly with Windows, they didn't when OS/2 was still alive, and someone expects Linux coders to do so for free..?!

To play devil's advocate, perhaps the real problem is that there are already user friendly Unix compatible environments : Windows and OS X. OS X is Unix, and can have X installed on it. Vista Ultimate can have the free Subsystem for Unix Applications (Interix) installed, the bundled NFS client, the free XMing X server, it's much easier to use software like tftpd32 rather than hack config files etc..

@stu

if you don't even know what a distro is, I wouldn't really be worrying about hardware compatibility at this stage!

Only just realised this?

$HOME, /usr/root/, bash, GUI, recompile, kernel, .conf, KDE, GNOME, XFCE.....

These are terms that users do not, should not and will not learn or recongnise. Users shouldn't need to wonder what desktop they are using. They should need to use a terminal EVER unless it's over the phone with guidence from tech support.

The Linux desktop is a joke. I hate Mac's, but I've got to hand it to Apple that they know how to make a good GUI. I personally feel that Windows is a good balance between technical ability and a usable GUI. None of them are perfect, but Linux is the least user friendly by far.

Config files, terminal and the such shouldn't ever be a concern of a user when doing something as trivial as installing Java or FF.

Fugly Linux Desktops

Yeah right, some of the best looking desktop are running on the linux platform.

Have a look at:

http://www.kde-look.org/

http://themes.freshmeat.net/

Sure, you will have to delve into how they did these desktops (and there is some dross in there as well as many diamonds).

But, don't think for one minute that X desktops are not capable of far more than other systems, it is one of the most flexible environments and one which has the most knick knacks and tools, you can do anything you like.

If someone wants to go Linux they can quite easily pay someone for an hours work on their desktop, and have something which is personalized to them.

One click install

Hey muppets who need dummy package install, Suse has 1 click install via YAST

linux drivers and linux kernal

It should be pointed out that Linux has a monolithic kernal rather than a microkernal like most other OS's. What that means is that most all the functionality of the kernal is in the kernal itself, including drivers. Whereas in a microkernal, only core OS functionality like file handling and process multitasking is handled by the kernal itself and all the rest of the functionality is handled by seperate utilitys. To get past this, there may be a need to fork a new version of LInux. Maybe some big companys like Sun and IBM could sort this out.

Linux works if you know how

So, if businesses want users to use linux then they should use opensource software as a base and produce an OS, it worked for Apple. Most of use kinda like the BSD base in it, makes it much easier to use and communicate with an Apple system.

But why on earth would anyone make it usable for the average user for nothing, it just sort of happens on some projects, but it should never be the goal. Unless they want to fork and flog it later. Commercial applications are designed for the user, open source is designed for the developer really.

The person who just wants to use new technology is well catered for by the various Linux distros, but if you want something specific then buy it or code it, asking others to code it for nothing doesn't work.

Hopefully Shuttleworth knows this, and will just hire the talent required, there is a big pool available, a lot of folks would jump off their project and go commercial with backing or even combine the two, there is loads of flexibility here.

@Mark Scott

"Hey muppets who need dummy package install, Suse has 1 click install via YAST"

No no no no NO! Most people are trying to escape Windoze in part because they want to choose what they're running, not have it served up to them by their distro vendor.

Take Eclipse - I need the full WTP-enabled, J2EE edition of Eclipse, with the CVS plug-in preinstalled etc. Ideally, I'd like the latest Ganymede release. The version served up by Fedora is a rebadged mish-mash of (I think) Europa, with no CVS, no server plugins etc.

On the Mac (or, dare I say it, Windows) I can just download the Ganymede J2EE fileset, unpack to a folder of my choosing and be off and running. No fiddling about putting it in /usr/local, creating links, then farfing about with the KDE menu system rather than click-and-dragging a shortcut ...

@AC

Such controversial comments, that you must hide behind AC..

If microkernels were at all effective in non-niche markets, it would see more popular support in mainstream OSS. This is *plainly* not the case. From what I've now (briefly) read about L4, its in one qualcomm handset, and makes up a lot of research projects at universities, where as monolithic/hybrid kernels can be found in virtually every computer system, from the embedded world to the mainframe.

If you take DragonflyBSD as an example, it is an incredibley ambitious project that quickly abandoned the micro kernel as a concept, even though it was one of the initial aims of the project. Microkernels are simply interesting research tools, that tell us hybrid kernels are the way forward, and that blindly sticking everything in kernel-space isn't ideal.

@ Tom

Your apparent need to use the fact that I value my online privacy as an argument against what I posted is an indication that you didn't find anything else to counter the point I made. You are making the same kind of mistake you did in your first post where you tried to use lack of knowledge of existence as an argument to support your opinion. Now, you offer low yield of material on a precursory glimpse as an argument for the same.

I repeat it once more for you: Just because you don't know of certain things doesn't mean they do not exist, likewise just because you don't find much information on something you didn't know anything about 15 minutes prior, doesn't mean that subject matter is dismissible.

If you are seriously interested in the subject matter, I suggest you study the academic papers by Jochen Liedtke, the researcher behind the L4 family of kernels. Until his untimely death in 2001 he had demonstrated a series of microkernels over several generations which were not beaten in performance by any of their contemporary monolithic kernels. Though it was a great loss for the microkernel research community to lose such a bright researcher, his legacy not only lives on but it is actively pursued with success. The L4 kernel was developed by those who continued after him and there are no significant differences in performance between L4 and modern monolithic kernels. To dismiss this outright simply because it isn't mainstream is just foolish.

Note that I didn't say microkernels are better than monolithic kernels, I simply stated that there is at least one microkernel which doesn't fit the description that this technology is outright inferior or obsolete. There are people who know more about kernel design than you and me who take this stuff seriously, so maybe you might want to consider taking it a bit more serious, too.

And yes indeed, I value my online privacy over all else. You obviously haven't had people stick their noses into your private life using the internet as a tool, because if you had, you wouldn't post as "Tom" whether it is your real name or not.

RE: Pretty os

OSX doesn't succeed because it is pretty. It succeeds because the system has been designed from the ground up to be well integrated and both logical (once you get used to it) and simple to use. In fact, it isn't pretty in all areas.

The whole enlightment desktop thing (although judging from the screenshots I have seen, I wouldn't say enlightenment is "beautiful" at all) shows one of the problems of Linux. Someone comes up with a new desktop, so what do the Linux bods do?

Do they ensure that the new desktop provides a consistant and easy to follow user interface? Err, no.

Do they work to ensure that apps will work without alteration on this new desktop? Err, no.

Do they work to ensure that apps have consistant user interfaces, with certain common operations in apps having the same menu options/shortcut keys (e.g. "Save" always being in the "File" menu and help accessible by pressing F1). Not in my experience.

Do they work to ensure that some of the more confusing aspects of Linux are hidden but still available to those who know where to look? Not judging by what I have seen.

No, they create a whole new distrubution of Ubuntu to run it. So, they add another Linux distribution to a market that's already flooded with multiple Linux distros and offers no indication to the non-technical newbie which they should chose.

I mention the non-technical newbie because this is the market that Linux has to penetrate if it is to succeed as a consumer OS. These people are not interested in the technical aspects of the OS. They just want a computer that they can operate fairly simply. They want something that will surf the web, play the odd video on Youtube, let them send email, do the household accounts and word process. They may also be the sort of people who are actually scared by options that look too technical.

You can moan about OSX and Windows all you want, but with both OSes, you can be fairly sure that if you download an app, it will work on your version of the OS, and will have at least some aspects of the user interface that you recognise.

This isn't an anti-linux post. I like Linux. The problem Linux has is that it is maintained by a load of geeks who seem to have little concept of what the average user needs from an OS.

Microsoft and Apple both have the advantage that they can afford to establish huge research labs specifically to design user interfaces and they can pour millions of pounds into this research. Linux doesn't really have that.

The other problem LInux has is that the only companies who do have capital to invest in it (Redhat, Novell, Canonical) who *could* be investing in UI design are either not interested (Step forward Redhat and Novell), or are totally misjudging people's needs (Canonical), and think that people want a "pretty" OS rather than a usable one.

However, if by "Beautiful", he means a well-integrated and easy to follow user interface, he is right. If he means the same as a lot of orther Linux users (a user interface with funky graphics and special fx), he is wrong. Funky graphics and special fx can look god, but after a few days of use, they can become an irritating waste of resources.

Modern microkernels significantly faster than monolithic kernels

"In the 1980s a lot of truly mediocre work was done on microkernels (which are similar to modern capability systems), and some equally bad analysis concluded that the problem was with microkernels in general rather than with the flaws of particular implementations. We now have examples of microkernels that are significantly faster than conventional operating systems, and at least one example of a capability system that is so."

emphasis on "examples" (in plural) and "significantly faster".

Jonathan Shapiro of Hopkins University in his essay on capabilities

@ Stuart Castle

"So, they add another Linux distribution to a market that's already flooded with multiple Linux distros ..."

You are confusing kernels with operating systems. Linux is a kernel. It is an open source kernel with a license that allows folks to use it and build an operating system around it. And that's what people do. Consequently there is no such thing as one single Linux OS. Every Linux distribution is a separate operating system, there are similar under the hood because they use the same kernel and a set of common APIs, still they are different operating systems and should be recognised as such.

Obviously, you couldn't do anything similar with Microsoft's kernel because it is a closed source kernel. However, Apple's kernel is open source, so you could do the same kind of thing the Linux distributions do: take the kernel and build your own OS around it. In fact, there are a few projects which have just done that. I remember one project which used the Darwin kernel together with a Gentoo derived userland.

You wouldn't expect a Darwin/Gentoo OS or any other such OS based on the Darwin kernel to be fully compatible with MacOSX, or would you?

The reason why there are so many Linux based operating systems is simply that the Linux kernel is very popular with independent builders of operating systems.

What else do you think you would be doing if you had a vision for a different kind of operating system? The fact that you want to do something different means you are in disagreement with the way many others do things. For some time you may try to convince those others to change things but often you reach a point where it is far more practical to start anew. So, you build your own operating system, probably using many of the components that others are using, but you will be doing things differently where your vision is different. If your vision is good and you are doing a good job turning that vision into reality, then you stand a chance that your operating system may set a new standard and perhaps other projects will then follow and adapt your ways of doing things. This is how open source works. The reasoning behind this is that competing ideas will lead to good ones to eventually survive and bad ones to eventually die. Like anything else, there are both advantages and disadvantages, the disadvantage being that evolutionary processes often take time, yet it may turn out to be more successful in the long (or very long) term, only time will tell.

@Tim Greve

"Thanks for suggesting likewise-open Darren. It just doesn't measure up to authconfig. To bad too. We are talking about 40k+ desktops here. It has to be simple enough for beginner tier one folks do handle without any real effort."

Why? If you've got that many desktops you should be administering them through puppet, not forcing users to muck about with GUI configurators. One of the really big advantages of Linux over Windows is Linux's low cost in systems administration when the number of deployed machines starts to get largish and you don't want to miss out on that. (40,000 desktops is only just entering the "large" space if you consider the number of machines Akamai and Google administer).

If you can't use puppet then at least control your site-specific configuration using the package manager. That is, you put as dependencies the software in your "standard operating environment" and you reach out and configure authentication, etc. Then as your environment changes you issue new packages and they are pulled in like any other software update. For example, we point our users to own own repo and ask them to install

example-all

which pulls in

example-software-list, example-software-authentication, example-configuration-authentication, example-security-configuration, example-corporate-software and so on. Some of these have dependencies, such as example-corporate-bookmarks. All these packages populate and configure the computer with the standard operating environment.

Users which need to do their own thing can easily blacklist any packages causing problems (which they usually don't, since we take some care not to be Nazis). Some groups issue their own packages containing their own software lists and configurations (eg, EDA or accounting setups). We host these on the same repo to minimise the amount of work for users.

One trick that's worked really well is to add a "live bookmark" (XML feed) for corporate notices and events. This is less intrusive and more interesting than company-wide e-mails.

@ daniel

"For headless servers, Linux has no equal"

That's an opinion and not everybody will agree with this opinion. I for one believe that BSD is superior as a server platform, then again, my buddy thinks Solaris has no equal as a server platform and another buddy thinks so about VMS. All opinions based on preferences, just don't sell them as facts like you did.

Microsoft should treat it as useful marketing

The corporate lawyers are no doubt flying on autopilot, but if this app is of decent quality then it's actually free viral advertising for Vista (and Microsoft have previously taken a completely different stance when people have replicated large gobs of the Vista UI online).

On the other hand I can easily see the Apple legal machine moving into action to smite him for the perceived desecration.