pouët.net

New hardware (read: driver support).

That's the only case I can think of, honestly. Because otherwise, the changes are pretty minor from an end user point of view.

So it's not that much of a choice right?

in soviet russia, operating system changes people!

knos: It's not like operating systems are static monoliths. They are constantly evolving, being updated with fixes, new features. That's why I always chuckle when someone's ranting about XP being "8 years old OMG! ancient software!"

Knos: what a load of rubbish.

Tomaes I agree generally. However it seems to me that what's most important (in the general case) about an operating system right now are its defects, rather than its features.

An operating system is a bit like basic, elementary furniture..

Only parts of an operating system, like the underlying architecture, IO management etc, but the UI and the way the user gets around and accomplishes tasks is also a part of the operating system -- at least by any reasonable definition.

Well .. is the UI really THAT different from version to version?

My stand point is that of a non-power user... I maybe use only 10% of the UI's functions..

but are you having fun with your linux from 2001? :P

Maali, are YOU having fun with an operating system?

Try installing windows 3.11 or even 95. You'll see why people end up switching to more recent versions :)

Operating systems aren't fun.

I tend to agree with knos, if I consider only the "typical" mass-market users. You know, the ones that use a computer for a purpose, to serve a goal. My mom doesn't upgrade until she has to, and I don't think she should either. She's usually not interested in having to learn the differences to be able to use the new features (that are usually not very useful for her).

However, this question was asked on a demoscene-forum, in a demoscene context (whoa, an on-topic thread on pouët? Someone, call the media!). And I think that calls for looking at this from a demoscene angle. We're technology-geeks. We love shiny new gadgets, why wouldn't we want to play with the new technology?

My post was supposed to read "Not only" btw, not just "only".


Not really relevant, because the statement was that the only reason to upgrade was to have better hardware support and that the changes otherwise were minor. So, if your mother doesn't want to learn the differences, clearly there are differences. :)

i have plenty of fun with the new improved solitaire in windows vista! and there's even poker in vista ultimate!!!!1

About the lowres demos problem, I think you should check the graphics card settings. I had the same problem in Asus EEEPC with WinXP with some intel gfx card and it was just a setting which I could also change in realtime with Ctrl+Alt+F11. Try this and if it's not the combination the see in graphic card settings. It solves the problem for most lowres software demos (accelerated demos was ok even in low res) except in very few ones.

I would argue that in terms of technological advancements, new mainstream OS releases are less interesting today than they have ever been.

There's lots of maintenance (mostly bugfixing, but I also count things like switching to a new driver model from time to time among this). Bugfixing and a lot of iteration certainly improves software for everyone involved, but I wouldn't count steady work on an existing codebase as a technological advancement unless there's significant breakthroughs in terms of quality, reliability or stability.

There's big changes in some isolated subsystems. For Vista for example, lots of work went into redesigns of the graphics and audio subsystems. While a big change "under the hood", from a user (or even developer) perspective they could've just as well been an iteration on their predecessors; they add some new functionality and package existing functionality differently, but I wouldn't describe any of it as game-changing; if anything, their designs are conservative in that they just took the existing programming model and adapted it to better fit to the actual hardware and software it works with. The UI saw a big makeover, but it's a reimplementation of the same concepts, not a new model of user interaction. Again, evolutionary, not revolutionary, improvement.

All of the grand plans that really would've made a notable impact didn't happen. We do not (or not yet) have semantically indexed file systems or networks. We do not (yet) have an OS based on .NET or a similar higher-level virtual platform (this would've really been a notable feat had it worked out, though I'm less than thrilled by it). OSes today iimplement nearly the same services they did in 1995, they do so in similar ways, using the same underlying models, and most of the APIs (on both the Unix and Windows sides) haven't changed at all! Of course, there's nothing wrong with this, but compare the differences between OSes in 1995 and OSes in 2010 with the differences between 1980 and 1995 or between 1980 and 1965 to get a sense of how much development has slowed in that area.

Of course, there is *some* interesting stuff happening: MS Singularity is a really interesting research project into a (so far) mostly uncharted area of the OS design space. zfs actually advanced the state of the art in storage, by addressing not only scalability, but also maintainability (and integrity) issues. There's tons of fascinating research on clusters and cloud computing.

But that's all very far away from the desktop OS space, whether it's Linux, OSX or Windows. It's really quite stagnant over here :).

Windows is a mature OS now. Any new releases are of course going to offer only minor improvements. Like the demoscene it's now less about the technology and more about the design.

Demoscene is still about cubes!

windows now is about png's

Oh, come ON! Did you really find MS DOS 6.0's addition of in-filesystem compression more exiting than the Win7 release? We quire recently got the first usable 64bit version of Windows, with the first non-single buffered GUI in Windows' history. DirectX 11 with it's compute shaders is around the corner only for those with Vista or 7. Times they are a-changin'! Screw DOS 2.0 through 6.22 and Win98 through WinXP.

ryg: the vista/w7 core is quite good (well, apart from the political decisions involved..wonder how the BSA determines the total number of installs of any given SW..) .

MS office section sucks, IMHO. :)

..ever noticed how the developer tools are so much better than the actual productivity tools which most of us are *forced* to use for everyday work (e.g. outlook, word, excel). I still shed a tear or two for VC6 (which was *really* good, in terms of resource usage/speed vs. features. and it was not written @MS but bought from an outside company..).

MS' plan of "devs, devs, devs" has really worked out for them, so far..

Is there any OSS IDE that is as good as/better than VisualStudio? Don't mention Eclipse, I think it sucks (ups..) (ok, reasons: resource usage, slowness, flexibility in terms of work flow (rigid directory structure, adding files from arbitrary locations is cumbersome).
Other OSS IDEs I have seen suffered from problems like for example being slower than their user when stepping through code (and yea, recent VS versions suffer from similar issues. gnarr.).

(ok, entering rant mode, better stop here. gn8 all but there's some truth to this)

"gn8"? :)

kusma, I never noticed nor even cared that the desktop was single buffered, honestly. Couldn't that be patched up with a new shell release rather than an complete OS revamp?

No, but I would argue that Windows NT was *much* more of a move forward in 1993 than Windows 7 is in 2009.

Windows 7 has yet another GUI design and lots of improvements all over the place, but is there even one piece in there that really offers something new?


XP Pro x64 was released in 2005. I'd argue it's the only usable 64bit Version of Windows so far.


Good for them, but excuse me whilst I yawn. You're actually trying to tell me that double-buffering GUIs was a *technological advancement* in 2006? With a straight face?


D3D11 is nice. As you just mentioned, it's not tied to the OS release. Compute shaders is one API, OpenCL is another, and there's also CUDA and BrookGPU. The latter two have been around from some time now. So again, not a piece of innovation, more like a must-have item they checked off their laundry list.

In D3D11, the one thing I'd pick in terms of new features is the new multithreading support that actually allows multiple threads to work simultaneously on independent fragments of a command buffer, and to submit draw calls for buffers that are still locked when DrawPrimitive is called but will be unlocked by the time the draw call is actually executed.

It's not a huge change if you just look at the API, but it replaces the underlying programming model completely - it's no longer a pipeline where you serially stuff things at one end, it's a network where a lot of work is done independently and then the sequencing is done "just in time".

So yeah, I'd agree that D3D11 is a technological advancement, but not for the same reasons you do.