OSNews

There are not really enough results yet to say this definitively but it is interesting that the graphic cards drivers are considered so popular. Is Desktop eyecandy really that important to everyone ?

The nVidia and ATi accelerated drivers are important for more than just "eyecandy" as you apparently are implying. Without them, 2D performance is fairly limiting. In fact, both of the drivers (at last check) also provided acceleration needed for video playback (dvd, etc.). Finally, some of us have work that actually requires 3D acceleration -- which has nothing to do with eyecandy or games.

I cannot comment on the specific things you do with video but it's not correct that you need anything on drivers to play-back video, DVD at all.

Well for things like MythTV with HDTV having the binary only drivers greatly reduces the load on your system for decoding. The same is true for DVDs etc.

Now don't get me wrong, if AMD/ATI open up some on their video cards my next gen card would be from them for sure. It's just that there really isn't much of an OSS alternative for real 3D performance drivers.

So does the radeon and nv driver. The ATI proprietary drivers aren't any better on video, and neither is nvidia's.

They're really pretty much only helpful for GL. Although, the radeon driver hasn't been working right on my dual monitor setup so I had to switch to the proprietary one (the free one made white horizontal lines when there were large numbers of changes on the screen).

And the Nvidia driver provides a lot better dual monitor support than nv+xinerama, but the radeon driver provides its own similar stuff so it's really in the same class for dual head.

So does the radeon and nv driver. The ATI proprietary drivers aren't any better on video, and neither is nvidia's.

I would like to see hard numbers proving that. My experience has always been much better with the proprietary ones.

"Is Desktop eyecandy really that important to everyone?"

Yes. People like pretty things.

I wouldn't say eye candy is all important for me, but in my experience the nvidia drivers are more reliable than the nv drivers and give me my favorite resolution of 1280*1024 which the nv drivers did not support, that's mainly why I use them. I don't really play games either (at least not under linux), but still the nvidia drivers are just about the first thing I install when (re-)installing a linux distro

The "nv" drivers work perfectly with SXGA resolutions. You are probably doing something wrong in the xorg.conf.

Actually, the nvidia drivers cause my laptop to freeze 2-3 times per day. Since I switched back to the nv drivers it's been solid as a rock. Yeah, glxgears and some of the screensavers play slower now, but they both work fine at 1600x1050 and I reboot for games anyway.

Is Desktop eyecandy really that important to everyone?

Its not just eye candy. There are many FOSS OpenGL games that cleary benefits from good OpenGL acceleration (jDoom, neverball, planet penguin racer [former tuxracer], cube, the patched SEGA genesis emulator generator-gtk comes to mind) and a pile of commercial ones (UT2004, Quake3/4, Neverwinter Nights, X2, Darwinia...).

Moreover, the multiple-screen extensions, HDTV resolutions or on-the-fly switching between different screen configurations are nice and progressing, sort of (these things are somewhat broken in different ways and combinations in each binary release from both companies as far as I know - try setting up a 1080i modeline with a 6200 card the latest nvidia driver and you'll [not] see).

I use my operating systems to get real work done.

Me too. About 8-9 hours a day. But I also use GNU/Linux for desktop/games machine and HDTV multimedia machine at home as well, so I need more than a database, text editor, ftp and lynx for my computing. I can see no reason not to expect or aim for this functionality when using FOSS.

That said, I think most of us FOSS buffs would benefit a lot from community driver improvements if the driver sources were released...

People want clear, crisp, legible easy-to-read displays. That's not eye candy.

Hardware-accelerated OpenGL makes software people use for productive purposes usable. Therefore people value high-quality drivers that permit them to use such software on Linux with their expensive accelerators. Maya, XSI, VeriCAD, and the like. Further the drivers appeal to people that like to play videogames in Linux either with native clients offered for id and Epic titles, or with Cedega. Then last you'll have interest in "eye-candy," but that's still a work in progress.

Firstly, the other options are mostly irrelevant. Things like Opera, Skype, Adobe Reader, Real Player, Nero Linux, Intel Compiler, Parallels and others have open-source equivalents which has resulted in VERY low popularity of these applications on GNU/Linux.

Out of those which do not (vmware, flash, graphics drivers, etc), I do believe nVidia/ATI drivers are the most essential. It's not just eye candy: it's getting the most out of the money you forked over for that fancy graphics card. Nv 2D performance is almost as sluggish as the non-existent 3D performance. Finally: what eye candy? It's difficult as hell to get Xgl to compile and run for any average user (or an advanced one like me: I chose to give up and wait until more work is done) and xcompmgr and friends have nasty screen drawing and performance bugs which leave them unsuitable for non-experimental use.

Regarding the point on video support: I don't know how good nv's Xvideo support is, but it's all hardware accelerated with the commercial drivers. Others have mentioned dual-screen (TwinView), overlays (Quadro) etc as reasons.

If you think that "desktop eye-candy" is all graphics cards are good for, and that anyone buying them is really just shallow you've got another thing coming.

It's vitally important that GPUs are available via drivers for software such 3D GPU assisted rendering, video compositing, 2d render acceleration, 3D CAD(D), 3D CAID among many other applications and uses that I've even bothered to get into.

Desktop eye-candy is pretty low on the list.

My bet would be on Photoshop. While there are some "alternatives", nothing beats Photoshop when it comes to serious business. I'm sorry, but Gimp is kinda lame, Pixel32 comes close, but I have a feeling I'm walking in a building made of cards (I did purchase it though).

Heh, I bought Pixel too, but at the RC stage it was so lame and buggy I haven't been back for an update since.

Gimp? Lame? Was that intentional?

If you look at the functionality, Gimp is great, but if you look at the UI, Gimp is just complete crap. It's almost unconscionable to present a user with such an UI sorry.

I like the GUI on the Gimp. I personally find it quicker and more intuitive than Photoshop. I prefer the separate window approach, I like the speed of the right-click menu. I frequently spend hours a day, professionally, working with the Gimp.

Your experience obviously differs - that's why there's choice, luckily.

"Sorry" but I've been using GIMP for years for web graphics, exclusively. I don't even have Photoshop. I don't understand / don't get what the big problem is with The GIMP's interface - at all. I have to wonder if people have even used this in the last 3 or 4 years.

We saved money on Photoshop licenses, which my team at work was demanding, and everyone whined and complained about The GIMP. 3 months later you didn't hear a peep out of everyone.

The GIMP's greatest sin is its interface isn't exactly the same as Photoshop.

As for print work, I never do that, so fine, it lacks those features (so I've read). But I've never needed to fire up another graphics application to do web graphics work, home photo retouching, and so on. Nor has anyone on my team at work.

People are going to be whining about GIMP 50 years from now no matter where it goes because of what its interface was in 2001.

It's got a main menu and a bunch of pull-down menus, cryptically titled "File", "Edit", "Select", "View", "Image", "Layers" and so on.

I don't get the issue. If people can't accept the fact that there are graphic applications that aren't exactly like Photoshop, that seems to be an inability or unwillingness to adapt.

Oh, and I came from Photoshop in Windows in the late 90s. I didn't get The GIMP at first either. I worked with it for a few weeks and from then on I haven't used anything else.

Not appropriate for every user, but certainly far more than the broken record complaining about its interface would indicate.

There are some crap applications for Linux. The GIMP is not one of them.

Well, I *do* graphics professionally for a living, and it's all 100% free and open-source software, including Gimp, Inkscape, POVray, Blender, and Xaos amongst many others. I tell clients that I wouldn't use an Adobe product if they paid me to, and after seeing my work, many of them say they can see why.

I'm sorry, but I've seen Photoshop user's skill levels, and it's nothing to write home about. There is an old saying that goes: "It is a poor craftsman who blames his tools."

I have to second the call for Endnote. OpenOffice really needs a serious bibliography tool. That will make a major impact on its adoption in academics.

The upcoming version of OpenOffice already includes EndNote-like functionality:

http://bibliographic.openoffice.org/

Including support for BibTeX, RIS, EndNote and MARC.

Bibliographic is vaporware at the moment. It is unlikely to be ready for OOo.org 3 at this pace of development.

For a good replacement for Autocad check bricscad http://www.bricscad.com/

It is a linux port of intilicad should be able to read all dwg, dwf files exported by Autocad.

For professional work there's not just Oracle.

You'll need good 3d-drivers for stuff like 3d-rendering or CAD. Both of there are used for very serious applications, like construction and simulation.
I believe this in fact is an important market for nvidia, and that's much of the reason why they keep going with Linux drivers.

Limewire is also Java based as well, it is simply just a matter of downloading the 'other platform' option on their website when downloading.

There is a free software Java implementation, does Limewire depend on Sun's Java?

IIRC, it doesn't really care; I've run it with the one included with MacOS X; the IBM JVM - so one would assume that if the said implementation conformed to the specifications, it should work without a hassle.

It's not actually needed - my laptop has an Intel chip and happily plays DVD's. It may make it a bit nicer, although I wasn't under the impression that my nvidia driver was actually accelerating my DVD's - could be wrong though.

True the drivers aren't needed but they help out a lot. In fact your Intel chip is accelerated because there are open drivers for that chip. Support for options like XVideo are important to a lot of us. My ATI card doesn't have enough RAM to support 3D but it does do XVideo and I wouldn't want to deal without it.

Skype? You know about the OSS WengoPhone? http://openwengo.com/

It's basically Skype, with the equivalent of SkypeOut - only the program itself is GPL'd. Too bad Skype has so much momentum. I prefer open solutions (which is why I recently dropped MSN Messenger (aMSN) in favour of Jabber).

I use Ekiga.
It's pretty nice. The great thing about having an open standard is that people get to choice the client they wish to use.

I have to find it a bit twisted that things like "closed formats" and "drivers" can ever be described as favorites.

People are simply *forced* to use them.

People are simply *forced* to use them.

True; but I'd put money on it; if there were two drivers for hardware, one opensource, one closed source, and both provvided equal features; you would probably find that the opensource driver would be more stable, more reliable and give better performance than the closed source one. The problem is, hardware companies would rather spread the myth that what they do is something voodoo and mystical and that if they opensourced their drivers their whole company would go bankrupt, which is a lie.

SUN whoses processors are based on SPARC specifications are full open and available, and yet, they are profitable; Broadcom make ethernet adapters, and they offer opensource drivers; same situation, they make a profit; VIA work with developers to support their hardware via opensource drivers, again, they're profitable.

Its a complete lie to say that if drivers were opensourced, companies would lose their competitiveness - as Nvidia and Ati try to make out in their justification.

>> "People are simply *forced* to use them."

Expain how that works. Who's going to come into my house and force me to run their software?

Let's be clear: When someone markets a closed-source product that does somethng that can't be done with an open-source counterpart, that's adding to the choices available to Linux userrs. If the closed-source software is tied to "closed-source" hardware, well, that's just one little piece of reality. A lot of people believe hardware companies should release their specs, but believing something is all in your head.

I do not agree that the free linux desktop does not exist or that the free linux desktop is a myth. My iBook G4 runs completely on free software. It does run a desktop and it must be usable, since otherwise I would not prefer it above OSX.

It is however correct to say that the free linux desktop does not support proprietary formats. So, no, in practice, I cannot use Flash and I cannot use Java using only free software (note that both are making progress, but still alpha or not even alpha, as you say). I also cannot access DRM-protected media. But still, I can access the web, I can read email, I can access nearly all media content with open source codecs and I can definitely read PDF documents. I also use the open source 3D drivers for the Radeon in my iBook. The lack of flash can be annoying, but for most of the sites that I visit, it is only used for ads, so I honestly don't care much.

I can agree 100% with you that you are better off with Windows, since you depend on several Windows applications and on several proprietary formats, but a free software desktop is definitely usable for a lot of people. And to answer your question: no one is saying you should be happy with a free software os... if you depend on non free software that runs better on a non free os, then just use that.

If you're ideological about free software, then you must be willing to make some sacrifices. Personally, I'm not ideological about it and I have no problem at all to use non free software and even pay for it. And I definitely have no problem using non free and commercial software on my free linux os. I do see the benefits of having free software, especially when you're more or less forced to use free software if you run linux on powerpc. However, the fact that I do run Linux, is because I really like the GNU/Linux OS (the kernel + the userland). It would be wrong to claim that whether Linux is free software or not does not play a role at all (I'm a programmer, so sometimes it does come in handy when you have access to the source), but if it wouldn't be good, I wouldn't use it. Fact is that I'm very happy with it.

I'm concerned with ideological subjects, but I cannot hide my head in the sand if I see a thing that even when I may dislike it, is true: a lot of widely used features are not working properly under a 100% free GNU/Linux desktop; for you it is Flash, for me another thing, and the so ...

Bah. Just because you're tied down to a bunch of non-free software doesn't mean the rest of us are. So maybe the Free desktop will never exist for you. My condolences.

You just have to take a second look at the list (again, not listing some applications or the missing ones) to see a lot of things that you can't do or at least can't do at its full potential with a 100% free GNU/Linux desktop ... and that's not my pleasure it being so, but it is the hard reality.

You just have to take a second look at the list (again, not listing some applications or the missing ones) to see a lot of things that you can't do or at least can't do at its full potential with a 100% free GNU/Linux desktop ... and that's not my pleasure it being so, but it is the hard reality.

Agreed. Like it or not, things are going to continue to become more and more proprietary.
If you continue to deny this fact, you will be like those Amiga or OS/2 holdouts who eventually get dragged away kicking and screaming. It's time to face reality, boys and girls .. open source is not going to revolutionize the desktop. You will always be two steps behind trying to keep up with proprietary apps and formats. I'm not saying this is a good thing, but seriously .. how long will you continue to struggle before you face the inevitable fact that 98% of the world doesn't give a sh*t about open source, and probably never will?

The key is not to struggle. :-)

If something can't be done on your platform of choice, use some other platform on a temporary or shared basis to perform that specific task, and continue to use your favorite one for everything else.

That's how I manage to keep on using OS/2 at home. If I don't have a DOS/Win16/native program that will do the trick under OS/2 and neither Odin nor Executor will run a Win32 or 68k Mac solution, I either use the Linux version via X on my Warp desktop (with the client running on another box) or flip to another Linux or Windows box via my KVM switch.

Networking can be your friend. Learn to use it. We aren't living in an age of isolated boxes anymore...

Edited 2006-07-26 14:30

98% may not care about what you call "open source", yet I know few people who don't use free software.

Linux is in what, its 15th year? FreeBSD is almost as old; its BSD roots older even than Linux.

They have prevailed, constantly, in spite of and in direct contradiction to the incessant allegations that this kind of development model cannot survive.

Yet we even - recently - have things like OpenSolaris.

People have been saying precisely what you are saying since the beginning of Linux - yet the desktop continues to improve, the development community grows, and we have more and more applications, and more mature applications with each year that passes.

The commercial world has not been able to stamp out free OSes or take it over with commercial "need-to-haves." This very poll is indicative of nothing more than there are fewer "must have" proprietary packages than ever before. The only two non-free apps I use are nvidia drivers and Opera (oh, and some codecs). That's it. Of the maybe 50 apps on my system, there are what - two proprietary products?

Your sentiments are very sort of 1996. I don't have any illusions that free software will eclipse the proprietary software world, but it will continue in parallel as an alternative to taking it in the wallet.

You just have to take a second look at the list (again, not listing some applications or the missing ones) to see a lot of things that you can't do or at least can't do at its full potential with a 100% free GNU/Linux desktop ... and that's not my pleasure it being so, but it is the hard reality.

It's not a hard reality, it's your opinion and/or unfortunate circumstance that you absolutely have to have some of these non-free apps. But that's not everyone's situation.

Of the apps in that list I have two installed. Adobe Reader and mp3 support. It wouldn't take too much effort to have pdf viewing and music entertainment at its full potential with another viewer and codec.

Okay, xpdf, kpdf, ... and ogg would serve you okay then, but you are happy then without being able to use your nVidia/ATI hardware at its full potential (if you own that hardware), or being unable to view some flash sites, and what about binary java applications or applets ? Free replaces for Java do not work 100% okay (in my own experience, and in the experience of some other persons). I think (it is my opinion, you're right at this one) that at least Java, flash and good hardware support are a must nowadays for a good general purpose desktop.

I mean, a GNU/Linux desktop works for me a lot of the times, even a 100% free one, but the fact that "a lot of times" does never become "always" in all of the years I'm working with GNU/Linux, just makes me pesimist and makes me think if someday it could be possible ... as I now see the market, only for internet serving (and for almost any serving purpose - though not for all of the ones) is GNU/Linux ready to compete with its rivals (mostly Windows and Solaris for my work).

Most of the times I had to deal with missing hardware drivers for Linux in the past (esp. video capture hardware) and present (those pesky wifi cards ... ) it boiled down to :

- Hardware vendor xy not willing to provide at least the specs so someone could write the driver for them (this happens to be most of the times gratis for the vendor, btw.)

- Existing standards for classes of hardwares are implemented in a messed up way ( think of the sometimes adventerous ACPI implementations for example)

- Device z is so rare or expensive, that it is difficult for developers to get their hands on at least
one exemplar for testing / developement / figuring things out.

- Legal hindrances (DMCA , patents, et al) that limit the possibilities of driver developers to reverse engineer drivers or introduce legal risks

and linear combinations of these issues.

So, for my naive view of the world, it is most of the times not a question of "do the devs consider a device to be mainstream enough to write a Linux driver / module for it" but rather if there are practical and legal possibilities to do so.

(Heck, there are linux drivers for rather obscure / non-mainstream devices in the vanilla kernel, being not mainstream enough is imho no sufficiant condition for not being included)

And yes, (even nonfinancial) goodwill from the hardware manufactor can help a lot.

regards

I'd add, in some cases lack of skilled manopower, especially if a device is severly bugged (even with a spec available) or complex to grasp.

Manufacturers often have a problem with a quality control of their hardware, so they attempt to fix bugs in drivers, but only in linux ones.

Proper way is to have vendors make minimally the initial functional driver, with full feature support. Either they or the communlity would then adopt it to be ready for mainline (it is preferable that vendor does it with some help from LKLM). Vendor should also set up dedicated bugzilla and even issue patches for problems which are discovered in hardware. They have much better testing capabilities than anyone else anyway. They should also follow kernel philosophy: try to reuse existing ABI's (and subsystems) and if necessary propose improvement to them if unsuitable for novel hardware (on time before product is repeased). With multiple incarnation of hardware (e.g. gfx cards) they should try to make a unified driver so that code is reused and code size & maintenance job reduced. They should step back from developing "custom" software solutions for their advertised special features and instead cooperate on shared framework, even if competitors can use it (bad examples: twinview vs. xinerama, winmodem software, etc.)