In order to get the next update of its Longhorn operating system out on time, Microsoft will delay some planned feature improvements, sources close to the company said Friday. Also, now that XP’s SP2 has been released, Microsoft is considering what other changes are needed to the XP operating system.
Microsoft overhauls Longhorn plans
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSAlert.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
They’re backporting Avalon?!
That puts the entire article in suspect… a backport of Avalon would require massive changes in the way Windows XP works right now, which is always why they said Avalon is going to be a Longhorn-only feature. I mean, if they backport Avalon, its just going to run on GDI+ – and there go your big performance benefits. That’s insane!
to what degree anti-trust, patent, and other non-technical concerns drive these ‘delays’.
In US Navy procurement circles, these delays are called ‘bow-waving’, the image being of a ship underway, and the requirements being pushed to the side in a surge of white foam, as the ship drives herself towards the deadline.
Here endeth the poetic imagery.
If they backport Avalon it won’t be a bad idea. AFAIK Avalon itself isn’t resource-intensive. Only the 3D, transparency and other effects will require faster hardware. They introduced the future Longhorn releases as separate ones. AFAIK there will be Longhorn with full of eyecandy and Longhorn for cheap hardware
Well, I am no Microsoft fan spending much of my time on various *nix’s. But they have said that they are considering more changes to Xp. This could also mean updated GDI that could work reasonably well with Avalon. Who knows? We certainly don’t. But it also sounds like M$ doesn’t either! Just my thougts.
going to be between longhorn and windows xp then? both will have avalon, both will have indigo. this was going to be microsoft’s attempt to really release something new and exciting now it looks to be little more the what XP was to 2000.
in other news: OS X will have most likely have spotlight a full year or longer before a beta of winfs is available.
i love microsoft roadmaps.
My thoughts exactly. With WinFS being completely absent and Avalon being backported, what possible motivation would people have to upgrade to Longhorn ?
All those features people have talking about for months and months have generated huge amounts of hype and hyperbole. Oh, they’re not going to be in the first release, but who gives a spit as long as they’re talking it up. There’s probably more people looking forward to the new Windows than the new “Star Wars”. Pre-order now, suckers.
1) Why can Apple convert from a GDI type environment to a fully 3D environments in an anuall .x release with no issues and MSFT with a much larger Developer staff can’t do it in a 4 year x. release?
2) Why are they dumping major features left and right? Longhorn was going to be XP+++ and now it looks like it’s going to be something like XP+. Also from the artical, it looks like you can add the + to XP with a down load.
3) Given 2; where LH = XP+ with Beta WinFS; why would you buy LH when you could just upgrade XP to XP+ and not have to worry about a beta file system?
4) Why can Apple go from HPFS+ to HPFS+ w/search in a .x release and MSFT can’t do it in a x. release?
Actually I can answer most of the Apple questions; Steve Jobs beleaves in under anoucing and over delivering. Put basicly, Apple waits until it’s ready to beta before anouncing what features will be in the version and they may include additional features in the shipping version. I would love to see MSFT adopt this methold!
Currently it looks like Apple is going to beat MSFT in shipping features again:
a) Avalon = Quarts Extreme; shipped 2 years before.
b) WinFS = HPFS+ w/search; shipped 1 year before.
c) Web-Services = WebObjects; shipped many years before.
d) 64bit; limited support now and full support next year. Still waiting for it to get out of limited beta (it currently doesn’t support most devices and has limited multi-media) on the MSFT side.
I mean, if they backport Avalon, its just going to run on GDI+ – and there go your big performance benefits.
Better running slowly than not at all. From what I’ve seen, even Longhorn is to scale back on the eye candy on less ambitious hardware. I recall that there will be three levels of eye candy (not named thusly): Win2k level, WinXP level and Avalon 3D fancy. There’s no reason that scaled down Avalon apps shouldn’t be able to run on Win2k and XP, especially since the scaling is done by the system.
The benefit to Microsoft is that developers will actually use the Avalon APIs if people can use them now, on their existing hardware/OS. How long did it take the theming APIs of XP to catch on? I still have programs that don’t use them. They should roll the new APIs into the standard .NET 2.0 runtime and let everyone use them – then everyone will.
Many investors have expressed concern about whether Microsoft can release new software fast enough to spur the company’s growth, as well as that of Microsoft-dependent technology companies. In the meantime, Linux providers and other companies with innovative technology, such as Google, are making inroads.
zdnet hits it on the head: providers with innovative technologies are making inroads. (implying microsoft does not have innovative technologies). i’m glad they snuck that in their. even the press knows microsoft doesn’t have innovative technologies and it looks like they think its going to hurt them. i think if they keep writing articles like this, it could come true (which i’m totally ok with).
They should roll the new APIs into the standard .NET 2.0 runtime and let everyone use them – then everyone will.
and all the developers that have learned and loved Windows.Forms will be pounding their fists! the second that happens Windows.Forms becomes deprecated, and basically obsolete.
ah well, that’s the way the cookie crumbles when you rely on a proprietary vendor with unclear roadmaps.
Darius: With WinFS being completely absent and Avalon being backported, what possible motivation would people have to upgrade to Longhorn ?
The article says WinFS in beta, not absent. Regardless, I would upgrade for the new interface built on those new features, the integration of those features, a more solid foundation for those features, etc. A backport may allow me to run Longhorn apps, but that doesn’t mean that they will run as well as they would on Longhorn.
From what I can see, the new WinFX APIs are better than their Win32 counterparts. Wouldn’t you want to promote their use?
Any of us who have been around long enough know this is business-as-usual for Microsoft.
We always hear about these amazing “innovations” and fantastic new features. Then they get delayed. Then they turn out to be less-than-expected. Then they turn out to have several flaws which require patches, updates, and service packs.
This is one reason I’ve already begun to transition my business to Linux and other open source technologies. They’re going places, progressing, and providing new opportunities.
I’ve been a Microsoft supporter, user, and developer for several years but can no longer use their products/technologies exclusively, I believe this is going to cost me my business in the end!
I’m going to just throw this out there and say that it looks like with all these massive mid-stream changes that either development of Longhorn has hardly gone anywhere, or development is running out of control in a bunch of different directions with no cohesion at all.
In my opinion Longhorn is quickly shaping up to be Apple’s Copland of yesteryear…. the question becomes who’s OS is Microsoft going to buy and renovate to keep up with the changing times?
-NA
>> “And why is there no editorial from you about the failure of Microsoft since they are taking so long to deliver?”
Maybe because Microsoft still holds a desktop monopoly and no one but OS enthusiasts have ever heard of Haiku or BeOS?
It doesn’t matter if Microsoft software is good or not, folks… it never really matter.
Brian Hawley
The article says WinFS in beta, not absent.
So are they going to ship a beta file system WITH the OS, or will it be available via a seperate download like the .NET framework is now?
Regardless, I would upgrade for the new interface built on those new features
Specifically, what new features are you referring to, and how do they directly benefit end users?
and all the developers that have learned and loved Windows.Forms will be pounding their fists! the second that happens Windows.Forms becomes deprecated, and basically obsolete.
Well that is one way to deal with the Mono/DotGNU incursion…
But seriously, Windows Forms will still run, even when they’re obsolete. And if Avalon is as drastically better an API than Windows Forms as it seems to be, developers may look forward to using it, and sooner than they would have otherwise.
What I would like to see is for the Avalon API to be finalized so that people can get to work on porting it to Cairo or some such for use by Mono and DotGNU.
I don’t see why we can’t get the full avalon experience on XP… MS allready talk about that some time ago.
Avalon can still show old gdi(+) apps, so it’s not going to be a problem here.
Brian: Regardless, I would upgrade for the new interface built on those new features
Darius: Specifically, what new features are you referring to, and how do they directly benefit end users?
The features I was referring to are Avalon, Indigo and such – you know, the WinFX APIs. They benefit developers. The developers then write programs, user interfaces and such that benefit users.
For some details on the user-visible benefits in the Longhorn user interface and applications, there have been many articles on the subject, some with screenshots, referenced on this site in the past several months.
Which OS?? probally Solaris, Modified wih those features.
It wouldn’t surprise me in anyway.
Microsoft, always touts features, pulls back half of them and then pushes them in as patches down the line.
I suspect that Microsoft is really dealing with a couple of things simultaneously. How to deliver an optimum number of features on a reduced timetable and PIRACY.
1. There is no doubt in my mind that Gates is weighing how many release of beta Longhorn to make available even as development continues in the background.
My suspicion is that in the end Longhorn will be considerably more difficult to pirate than XP.
2. Which of the three big enhancements to focus on. I suspect concentration will be given to WinFS and Avalon with Microsoft relying heavily on further enhancements to .NET to carry it for a while.
3. To upgrade or not to ugprade. Gates has to decide what kind of upgrade path from XP to Longhorn will be possible. I suspect he’s still on the fence at this point.
So, in the end, I think reducing this argument to a simples features consideration is probably oversimplying the actual picture. But hey, I’m just a programmer.
I think that MS, for hype’s sake, came out and said: “This is what we will do in 2006, look how cool we are”.
The OSS community IMO has been taking this as a goal for themselves: to have something that competes with Longhorn in 2006.
So, in 2006, a GNU/Linux system will be comprised of the following (excuse the projected version numbers):
Kernel 3.0 (likely)
GNOME 3.0 (potentially completely svg-based desktop capabilities)
KDE 4.0
X.org modular release (3d accelerated across the board with compositing, real translucency, cairo, glitz, etc.)
A complete rewrite of GTK (supposedly incredible speed gains)
Plus a more organized freedesktop.org set of applications and standards to help integrate applications across the board without sacrificing choice. The potential is incredible.
All this, and now Longhorn each month seems to be pushed back or scaled back. MS should be prepared.
The features I was referring to are Avalon, Indigo and such – you know, the WinFX APIs. They benefit developers. The developers then write programs, user interfaces and such that benefit users.
From what I have observed, .NET benefits users about as much as Java does – developers will now write apps that are at least 3x slower than equivalent VB, Delphi, or C++ apps. So why do I care if they are easier for you to write?
As a developer, I am excited about WinFX. As a user, I don’t really give a rat’s ass.
For some details on the user-visible benefits in the Longhorn user interface and applications, there have been many articles on the subject, some with screenshots, referenced on this site in the past several months.
Would you care to name a few here?
1. to answer question number one Apple didn’t release their equivalent to replacing GDI (Quartz) with a .x release of OSX. They added it when they made the first version of OSX and the equvialent to GDI was in OS9 (Quickdraw Toolbox).
Does anyone know if the WinFX API is planned to be implemented in a manner similar to .Net was with Mono? WinFX is simply an API, not a suite of programming languages, like .NET, but would a Linux port be possible?
Also, windows.forms deprecated? How does this hurt Mono/DotGNU? Is windows.forms what is, in fact, implemented in Mono? I think I’m missing something…
Certainly MS is not doing away with .NET…right? Mono will still be useful for writing and running .NET apps, run???
If anyone would be so kind as to provide a little insight, I’d be grateful…
Backport to XP, hmmmm. Looks like all XP users are going to wind-up being beta testers. I wonder when their will be a service pack for Avalon and Indigo.
For the longest time I was serious disappointed that MS would relegate w2k to a patched system. The reason is that w2k users had to go out and buy w2k and xp users had it preinstalled on their systems.
Who didn’t see this coming? Its Microsoft’s bad habit of overpromising and underdelivering once again rearing its ugly head.
This is why it’s a bad idea to go 4 years between major releases, and try to shove tons of new features into a single release. This is why Windows 2000, which was perhaps Microsoft’s least ambitious NT release, was also it’s most successful one. It’s far better to make incremental releases, with one feature at a time, than to try to make one monster release with everything and the kitchen sink. Apple has proven itself to be very successful at this. 10.0 was released with a lot of things missing. As time progressed, cool features were added, gradually, to the base system.
In any case, the worse the final Longhorn release ends up being, the better it is for Linux and Mac folks. A 2007 timescale release of Longhorn (or an earlier release, with all the features not coming in until later), would ensure that Apple and *NIX will have the nifty Longhorn features before Longhorn does.
This will give OS X and Linux the spot light for a few years to come, by that time the 2 OS’s will already have more plans in the work. Apple already has former BeOS people working for them, and the talent behind Linux makes MS look like they are currently scratching thier heads. I think MS is getting to big to move rapidly now that it can’t steal or monopolies like they did in the past. Come on MS innovate! You have the talent, get’em working!
Joe P, actually for quartz Extreme I have discs here from OS X 10.2 dated Summer of 2002. It’s more than 2 years ahead of what Microsoft has in store, Panther was the first OS Apple shipped with Quartz Extreme packaged.
As far as the computer industry goes this being 2004 this is old technology.
I cannot understand how a company with such resources available cannot deliver. I presume that the technology that they announced is not beyond the realms of possibility. Mac OSX has had, or is about to release variants of the technology already. Linux is getting there as well. However, MS seems incapable of delivering it. I just don’t understand how that can be. What is going on at MS that is having this result?
A cynic would say that it was vapourware from the outset. Hyped simply for the purpose of trying to keep people from switching to OSX or Linux. But given that MS must have known that they can’t kill OSX or Linux with vapourware, they wouldn’t have risked the reputational disaster of not ultimately delivering.
Is it that the cathedral development method is a disaster? It can’t be, given the ability OSX has to deliver high quality features at the pace it does. Although. maybe it’s a question of scale. Maybe, the cathedral method only scales to a certain point (in terms of number of developers). Once you get beyond a certain point, in terms of numbers, the only development method that works is the bazzare method.
The final option is that MS is just incompetent. I have to say that there is ample evidence for this conclusion.
However, whatever the answer, this is good news for alternative platforms, and I pity those poor people who chose software assurance. At our firm, we were offered this, and IT people were pushing this hard (MS is all they knew, and the MS marketing people had shiny trinkets to give away). However, thankfully management had a clue, and were able to remember the words of the great Chuck D (or was it Flava Flav):
Don’t, Don’t, Don’t, Don’t believe the hype.
Matt
And get a good stable product. Look at Linux, with every upgrade for Linux I always have something that goes wrong and i attribute that to sometimes speedy is not necessarily good. As for Avalon and Indigo Im all for them being backported to Windows XP and Server 2003 and this isnt the first time Microsoft has said that they are going to make Longhorn components available now for XP and Server 2003, look at MSH (Microsoft Shell) that was supposed to be a Longhorn only feature yet they are now going to release it with .NET Framework 2 because they didnt realize just how popular MSH was going to be until they opened it up for Beta testing and that has been one of the key requests. Backporting is actually a good thing because I know more customers that do love Windows XP than those that hate it. Backporting and cloning is not a bad thing. Im fully expecting someone to clone Spotlight for earlier versions of the Mac OS and probably Windows.
Maybe Microsoft has all these problems because they must maintain backwards compatibility…. with several thousands of potential configurations that may break if upgraded. Mac OS X made it clear that it is very different from OS 9. In Linux, things break, and we take it for granted. They probably have to spend more time ensuring older configurations do not break than in developing the new stuff.
Well, you have to consider that Quartz “Extreme” as of 10.2 is much less impressive than what people make it out to be. It’s not a “full 3D environment” like people claim. Rather, it’s a simple application of OpenGL to accelerate the compositing stage of the UI (which is at the very end of the pipeline). It’s not comparable in technical complexity to what will be in Longhorn, which is accelerating actual drawing with OpenGL. That brings into play all sorts of complexities, like sharing of OpenGL contexts, modification of driver APIs, meshing a 2D Postscript model with the 3D OpenGL model, etc, etc, that Quartz “Extreme” doesn’t have to deal with at all.
AFAIK the main speed benefit of Avalon is in the fact that it sends most work over to the graphics chip, and it’s gonna be tough to overload one while simply running desktop graphics so performance issues would probably be less noticeable (except of course the infamous lack of a refresh on long processes, *cough* RedHat’s up2date *cough*).
With regards to Apple’s development model vs Microsoft’s, I think you’ve got two things in play. First, Apple has a smaller, more agile development team. It’s easier to get them to do things than to get the thousands of Microsoft employees to coordinate. Second, you’ve got the fact that Apple simply has less work to do. Consider all the things that MS needs to reinvent, and Apple does not: an advanced language like C# (Apple already has Objective C++); a modern API like .NET (Apple already has Cocoa); a set of media APIs like XNA (Apple already has OpenGL/OpenAL, and can use OpenML when it comes out). By taking advantage of open standards and open implementations, Apple can get by with it’s smaller development team, and get the agility advantages that a smaller team carries.
Now, in comparison to *NIX, Microsoft has to deal with it’s more monolithic nature compared to Linux’s distributed nature. *NIX has very cleanly specified and formalized interfaces between it’s various components. The window system is decoupled from the toolkits via the X protocol. The kernel is decoupled from the GUIs via POSIX. Because these standards are relatively static, each side can change rapidly without having to coordinate as a whole.
I don’t know why everyone’s saying this is a bad move for Microsoft.
Apple under-hype and over-deliver and look where it’s got them – second place and sliding, in spite of having a great product. iPod got hyped to hell and now everyone has $300 worth of easily steal-able walkman in their pocket.
Linux always hypes up new (and usually arcane) features within the development community, but the distributors NEVER talk-up their up-coming releases.
GMail’s getting pre-hyped by all the blogheads out there, who really need 1GB of e-mail everywhere they go but don’t know how to use a DVD burner.
Microsoft did it with Windows 95. A long pre-release hype period followed by a huge ad campaign, followed by a novel but frustrating first release. They kept tacking on features like DirectX and IE, and patching bugs. Drove everyone mad until 98SE smoked out most of the biggest bugs. And yet everyone went nuts for XP. And they’re already going nuts for Longhorn. You just know everyone’s boss will buy it, so then we all have to have a copy at home. A tie-in with some big game publishers ensures little jimmy wants one all his own. New APIs and IDEs and other TLAs keep developers from seeing the wood for the trees. And everyone gets locked in for another half a decade… Feature slide is just good business sense.
Some of you Linux users need to get a clue. WinFX(Avalon) API set goes far beyond Glitz/Cairo or the latest freedesktop buzzword that you are co-opting(not even Quartz can come close). They use the OpenGL to help speed up the rendering of the 2D user interface, thats it, thats all, just like Apple’s Quartz and that been out for 3 years+. Worse yet, many Linux users proceed to bitch about Microsoft being behind the interface curve without looking at thy ownself. WinFX will be creating windows in a 3D user space. Each window in Longhorn will be created with polygons, and lots of them. Each window element(buttons, icons, etc.) will be made up of at least 1 or more polygons. For a example, take a look at Microsoft’s Photo Triage(demoed early this year), each and every picture in the main window of the program is a polygon using a image as a texture.
Quartz, Glitz/Cairo, are not the same as Avalon.
Sun’s Looking Glass is the closest to Avalon.
As for WinFS, I always thought this one would bite them in the ass, and it did. It’s the sole reason why Longhorn is behind. If it wern’t for their foolish agressiveness in getting WinFS into next Windows release, Longhorn would be out Late 2005 with Avalon and Indigo. WinFS should have been scaled up(like what Apple’s doing with spotlight) rather throw everything in at one shot.
The backporting thing is bullshit too(hype and spin to keep the investors at bay). The WinFX/Avalon API set is just to advanced for XP.
I think you underestimate what Linux, particularly Gnome is going to be doing. With future releases of gtk everything will be drawn as an svg, the svgs will be rendered with cairo, which will be using a open gl backend, with a compositing manager using the same things, everything will be rendered on screen with OpenGL. I don’t see any reason why the same things (every image in a photo manager, every icon in a file manager, etc as a seperate texture) couldn’t be done.
Surely an all svg interface, with everything rendered on screen with OpenGL is going to be just as advanced, as Longhorn’s DirectX rendered interface.
(How can linux users co-opt freedesktop buzzwords?)
If I’m wrong about the scope of the svg/cairo/glitz/composite possibilities, somebody please tell me.
I don’t think you’ve got the foggiest idea what you’re talking about. Yes, Longhorn is more advanced than Quartz “Extreme”. No, Longhorn is not fundementally different from Composite and Cairo/Glitz. How do you think Glitz uses OpenGL to accelerated 2D rendering? By tesselating 2D shapes into triangles and quads! If you want a 2D view, draw those in an orthographic projection. If you want a 3D view, just switch to a perspective projection. How do you think an OpenGL composite manager will work? By using the windows buffers it gets from XComposite as source textures for quads! Wanna rotate that Mozilla window? Just switch to a perspective view, and draw the quad at an angle! Since a glitz canvas is fundementally an OpenGL context, you can do anything in it that you can do in OpenGL. Yes, that even means filling Cairo shapes with complex pixel shaders, if you are so inclined.
Now, Longhorn does have some nifty features like a retained-mode canvas (like Evas), and animation features, but if that is needed, those can be added in time. Fundementally, however, the two systems are pretty well-matched in capability.
GTK+ will not render everything via SVG. Rather, the idea is to move rendering over to Cairo.
I mean, if they can backport the rich API of Avalon, XAML, and still tie it into the powerful 3d accelerated hardware rendering they had planned on it for Longhorn, then everything’s going gravy – I’d much prefer it this way.But (grammar alert!) if they could have done this all along, why wasn’t this the plan from the start? Is it really less work to go to town on the Windows XP internals than delay Longhorn by a few months? (Longhorn will, after all, still have Avalon, and they will both be released at the very same time, claimeth the article).
WinFS being in beta form is really the only disappointing note of this article, if its true – I really, really want metadata. I want it now, too; having it deployed at the very release of Longhorn means it will almost instantly be deployed by Linux and other platforms in a matter of days afterwords, and there will be a lag time of possibly days before someone figures out how to convert WinFS metadata to a more “open” format.*
*I’m one of those big assholes who consider a computer a tool instead of a philosophic soapbox
“glitz is an OpenGL 2D graphics library and a backend for gl output in cairo.” Moving GTK+ to cairo rendering is definitely not incompatible with OpenGL hardware rendering rendering/transforming.
@Phil: I never said it was. The point is that it’ll GTK -> Cairo -> OpenGL, not GTK -> SVG -> OpenGL.
@Richard S: Huh?
Some of you Linux users need to get a clue. WinFX(Avalon) API set goes far beyond Glitz/Cairo or the latest freedesktop buzzword that you are co-opting(not even Quartz can come close).
I can’t help but think this is directed at me, since I was the one who brought those specifics up, originally.
I have to say, you are not judging my comment (at leasT) correctly.
The point is that in two 18 months, Linux will, incorporating the elements I mentioned, have an incredible desktop potential. Distro releases will be awesome. Linux will also have ReiserFS 4 with meta-data and hopefully FM’s to take advantage of it…this coupled with all of Linux’s inherent advantages.
So, I ask you, what will Windows’ true advantage be at that point? POSSIBLY, if it is indeed better, Avalon. Possibly.
Linux, IMO, is already VASTLY superior to Windows, while I know It’s not for all…
Have fun.
I think Glitz/Cairo isn’t really comprable to Avalon. Cairo is an SVG rendering library. Glitz is a backend, allowing Cairo SVGs to be rendered using OpenGL. This is more advanced than just Looking Glass or Avalon. The latter are still resolution dependent.
Once Glitz becomes the defacto on the linux desktop, the desktop will become resolution indepentdent. You’ll be able to have high resolutions without text and stuff getting smaller.
WinFS being in beta form is really the only disappointing note of this article, if its true – I really, really want metadata. I want it now, too; having it deployed at the very release of Longhorn means it will almost instantly be deployed by Linux and other platforms in a matter of days afterwords, and there will be a lag time of possibly days before someone figures out how to convert WinFS metadata to a more “open” format.*
Just to let you know, FS’s like Reiser 4 already support metadata today. Reiser 4 stabilized just last week, so soon apps should come out that utilize it.
Sorry to break your little heart.
Cairo is not an SVG library. The two are completely different. SVG is retained-mode, Cairo is immediate-mode. SVG is XML-based, Cairo is a C library. Cairo doesn’t even use the SVG imaging model — it uses the PDF 1.4 imaging model. Cairo also won’t be more advanced than Avalon — both will be resolution independent.
PS> If you’re text is getting smaller when you increase your resolution, then yell at whoever makes your OS for not using the DPI setting properly.
Cairo can render SVGs in hardware, but not everything rendered by Cairo is an svg.
BTW Wasn’t Cairo the original code name for Windows 95, that was supposed to have most of these same sort of features that Longhorn is (or was) supposed to have?
(forgot to put my name in that earlier post)
I’m curious to see what they do with it. WinFS is going to be beta when Longhorn is released. Likely those who initially switch won’t bother with WinFS. So Microsoft sends them a message a few months later:
You are using an outdated file system, please burn this following file to a floppy. Then reinstall your system using the floppy so you can switch your file system.
Cairo is the codename for NT 5.0, which became Windows 2000. Cairo was supposed to have an advanced OO UI and an object-oriented filesystem. Basically, like Longhorn, Cairo was supposed to have all the buzzwords that were in vogue at the time.
I hate to say it, because it’s already 90+ degrees (F) outside and a flame war won’t help any, but I have a funny feeling MS’s Longhorn problems are due more to a difficulty in migrating to 64-bit processors than anything else. This is understandable, because that area simply hasn’t standardized enough for MS’s “one size fits all” approach to computing. (Yes, MS has “HPC” ambitions, but that’s a different market than the home/S-M-Business arena.)
Linux (yes I’m biased) has an advantage in this regard, simply because the OSS development model tends to be quite faster on it’s toes. Even inside MS, only a handful of developers understand the majority of Windows architecture (by necessity, having too many people know the entire system’s internals is dangerous to MS’s way of doing business), so bug stomping is necessarily more limited.
If MS releases too early (in this case, before there is a clear consensus on processor style), they would have to do massive development to adopt, and major spin control as to why MS is falling behind in the market, or at least not selling their new OS at expected rates. While the items in the posts above no doubt have their effect, I think this is the more problematic item for MS.
It’s easier, if more dangerous, to wait until this sorts itself out, then release accordingly. I also think this is why MS is trying to play AMD and Intel against each other (every couple months, one is “favored” over the other by MS execs), in order to force some kind of agreement earlier than it might develop otherwise.
Just my opinion.
Er… meta-data in the file-system? No thanks.
I was referring to something like gnome-vfs or storage, beagle, wish they’d just pick one and stick to it.
Rayiner Hashem everytime you write something I learn more…..keep writing! Ken
Home page for Cairo:
http://www.freedesktop.org/Cairo/Home
Home page for glitz:
http://www.freedesktop.org/Software/glitz
Cairo/glitz has nothing to do with SVG.
But there is an SVG implementation on top of cairo, libsvg-cairo
http://cairographics.org/libsvg-cairo
Cairo has several backends:
glitz – OpenGL
PDF for printing
xlib
bitmaps, PNG, etc
Cairo is a 2D only API. When Cairo/glitz is drawing to OpenGL you can’t get to the third dimension. If you want to draw 3D ignore Cairo and call OpenGL directly.
X on GL is mainly waiting on superbuffer/pbuffer support in Mesa solo. It’s being worked on. pbuffers are needed for each app to draw into, these buffers are then composed into the final display buffer.
What is the purpose of using the glitz (opengl) backend to cairo to draw 2d enviroments?
It appears the vapor wear was vapor wear!
Drawing the same 2D drawing with glitz is about 100:1 faster than xlib. glitz uses the graphics hardware to draw. Just compare the memory bandwidth, a PC may have 1GB/s to AGP while the GPU may get 24GB/s.
Fundementally, there is no reason why you couldn’t draw in 3D space with Cairo, right? It should just be a matter of setting up the viewport properly. It might be useful for certain cheezy special effects.
@Anonymous: The point of using OpenGL as a Cairo backend is to allow acceleration of the rendering code. The current glitz allows you to offload rasterization, gradeint-fills, transparency, and anti-aliasing to the GPU. If the next generation of shaders allow you to generate and kill vertices, you could also accelerate tesselation and evaluation of bezier curve vertices.
I think you would need to hack Cairo in order to draw to a different transform. It’s not hard but Cairo doesn’t give you the API to do it. Of course you can use Cairo to draw a 2D image to a pbuffer. Then use OpenGL code to manipulate it.
Here’s the usenix glitz paper. It has some benchmarks in it.
http://www.cs.umu.se/~c99drn/opengl_freenix04.pdf
this makes sense if they introduce these backends now more apps will support it when longhorn is realesed. and the apps that support that backend will probly run better in longhorn
so there will be more reasons to upgrade.
if they didnt do it this way there wouldnt be many longhorn redy apps when longhorn is realesed and users wouldnt have any big reasons to switch
The point is that in two 18 months, Linux will, incorporating the elements I mentioned, have an incredible desktop potential. Distro releases will be awesome. Linux will also have ReiserFS 4 with meta-data and hopefully FM’s to take advantage of it…this coupled with all of Linux’s inherent advantages.
So, I ask you, what will Windows’ true advantage be at that point? POSSIBLY, if it is indeed better, Avalon. Possibly.
First of all, let me say that I think there is a huge difference in perspective between Linux and Windows users when it comes to OS upgrades. It seems you guys view distro upgrades like they were the second coming of Christ, especially if there’s a new KDE/Gnome or kernel upgrade in there. I mean, file systems, kernel upgrades, new graphics rendering .. you guys actually get excited about that stuff.
For us Windows users (most of us anyway), a new version of Windows is just an upgrade. Hell, I haven’t been excited about a new version of Windows since Win95 was released way back when. To be honest with you, I don’t give a rat’s ass when Longhorn comes out. Why?
Look at it this way .. even if Linux manages to add all the features talked about in these comments and fixes some of the usability issues that have been plaguing it forever and a day, it’ll be an awesome desktop operating system that is still running the same lame-ass applications, albeit 2 or 3 years more mature. (Which means that hopefully, 90% of the project on SourceForge will be out of the alpha stage.)
So while Windows (the OS) pretty much remains stagnent with releases every few years, the apps (and I’m talking 3rd party here) are always evolving and will continue to do so at the same rate they always have. For that reason, since some of the most critical apps I use currently offer 10x what anything in Linux has to offer, and all the ways that Linux is supposedly going to be ‘catching up’ to Windows is in the OS department, I couldn’t care less if they delayed Longhorn until 2038.
As an end user, I don’t care about Avalon, about Indigo, about Glitz, about Mono, .NET, Java, KDE, Ghone, WinFX, WinFS … in fact, you can take all of these technologies and shove ’em up your ass for all I care. The apps are what are important to me – they are the only thing that matters to me. Hell, give me superior apps, and I’ll run whatever opearing system you want. Hell, I’ll even install The Hurd if I have to.
With WinFS being completely absent and Avalon being backported, what possible motivation would people have to upgrade to Longhorn ?
There has been no business reason to upgrade windows since NT4. There is simply too little increase in productivity to make it worh while, only lots of costs, e.g. for retraining, loss of productivity during install, licencing,..
The only reason to upgrade today, is that at regular intervals Microsoft drops support, and you have to have a newer version to get security and bug fixes.
The apps are what are important to me – they are the only thing that matters to me. Hell, give me superior apps, and I’ll run whatever opearing system you want.
Give us examples. What sort of apps do you use, and which programs do you consider to be best in class?
“This is why it’s a bad idea to go 4 years between major releases, and try to shove tons of new features into a single release. This is why Windows 2000, which was perhaps Microsoft’s least ambitious NT release, was also it’s most successful one”
You’re off with the fairies, my uninformed little friend.
Windows 2000 was a major overhaul of Windows NT, especially with the addition of Plug and Play, Active Directory etc.
And there was about 4 years between NT 4 and 2000 (NT4 released mid 1996, 2000 released [oddly enough] early 2000).
One question, Microsoft used to name all their development versions, standards and such (Joliet, Chicago, Cairo, etc) after towns in Illinios, so wouldn’t Win 2000s code name have been pronounced KayRo?
Why’d they change their codenaming scheme anyway?
Ever think they thought Egypt that time??
Windows 2000 was definitely an unambitious release:
1) It’s major improvements were DirectX and updates to the driver API (WDM, Plug & Play, ACPI). All these features were already in Windows 98, it was a matter of moving the code over to the NT kernel.
2) It was much less of an overhaul than NT 4.0 (totally new GUI, kernel-mode GDI, large kernel structural changes), and XP (new GUI, first home version of NT, etc).
3) It was four years in the making, for a very incremental feature update, instead of the massive overhauls that Cairo was supposed to have.
I don’t know, but when they have so many other place names that happen to be in Illinois as code names/standard names etc (Rockridge is also a town in Illinois, but thats not really a MS standard), you have to wonder, how was it pronounced internally? And did they use it because its a place in Illinois?
Whats with all the Illinois themed names of software/standards anyway?
I would hate to have been one of those many companies who bought the upgrade assurance plans on the promise that updates would come every three years.
MS is going to have a hard time getting those companies to renew those service contracts. (It’s the company’s fault in the first place giving MS money for nothing but promises. Suckers.)
I can clearly understand why MS cuts down some feature. I you look to the past Windows 2000 was in beta mode for a about a year. An in a beta stage software you shouldn’t have to implement new features rather you are going just to fix remaining errors.
Windows is an oversized OS so they need a lot of time to test.
Anyway a beta should be here by early next year. Not much time to implement new stuff.
“Windows 2000 was definitely an unambitious release:”
Well that’s the first time I have ever heard of Active Directory being “unambitious”. It’s damn close to the only reason why MS has a popular server platform at this point.
To call it an “unambitious release” would be a huge stretch even for lesser imaginations. If not for AD; I could easily see MS out of the server business entirely. It was what they needed, when they needed it.
What AD has evolved into with 2003 is mighty damn rich, and works VERY well in the right hands.
Rayiner, Where did you get your information? Because I don’t see how one can switch from 2D ortho mode to a 3D projection matrix. Afaik, in ortho there is no depth. So switching to 3D would leave with a 2d plane of windows; it would look silly.
Also, someone mentioned that CPU to AGP is 1GB/s but the GPU can handle 24GB/s. How will you ever use that 24GB/s if you can’t send data to the GPU Fast enough? The GPU was not build for things like 2D desktops. You will most likely still need to provide it pixel information that has been processed by the CPU.
A lot of people got very excited about Avalon, but I’m 100% sure that Avalon will not be all that earth shaking. I doubt it’ll even be more advanced than Quartz Extreme, not that QE is by any means really advanced (just the most advanced of this time).
It’s really cute to be speculating, and calling out the winners so soon. But I will just sit back and see what happens. So far I’m liking what I see with Xorg and stuff, and I am putting most of my hopes onto Linux.
I was hoping longhorn would really innovate.
If features like Avalon did make it in, how would KDE/Qt compete. I already know what the plan is with GTK+ and Cairo, will Qt use a similar method?
Also, someone mentioned that CPU to AGP is 1GB/s but the GPU can handle 24GB/s. How will you ever use that 24GB/s if you can’t send data to the GPU Fast enough?
I see why you aren’t writing graphics drivers. You get the 24GB/s when things are already loaded into VRAM. That’s why games load all of their textures onto the card. The GPU then composes these textures together to build each frame. A small amount of info is sent across AGP each frame telling the GPU how to combine things. In the 2D case fonts and bitmaps are preloaded. Fills and gradients run locally on the GPU too. The biggie is the final composition step where app pbuffers get combined into the display buffer. That last step may involve alpha blending 20MB of data.
There is no guessing here. Look at the glitz usenix paper, 2D glitz on GL is 100:1 faster than direct 2D mode. MS and Apple see this too, they are also implementing 2D drivers using 3D mode.
Well that’s the first time I have ever heard of Active Directory being “unambitious”. It’s damn close to the only reason why MS has a popular server platform at this point.
To call it an “unambitious release” would be a huge stretch even for lesser imaginations. If not for AD; I could easily see MS out of the server business entirely. It was what they needed, when they needed it.
What AD has evolved into with 2003 is mighty damn rich, and works VERY well in the right hands.
You can count the number of people who have rolled out Active Directory, fully, organization wide, on one hand. The people who rolled out Windows 2000 did so because it is Microsoft, and because they were told that it would be the OS for the next ten years – and they believed it.
XP sp2 was a big flop it introduced more bugs and holes in windows and made .Net crash more often. Windows & .net is still very flacky … marking longhorn wont fix the issue.
“Windows is an oversized OS so they need a lot of time to test.”
Now I’m lost, how is it oversized? Are you referring to the variety of applications it ships with, or that NT was originally intended to use a Microkernel and now doesn’t even come close to one?
The point is that in two 18 months, Linux will, incorporating the elements I mentioned, have an incredible desktop potential. Distro releases will be awesome. Linux will also have ReiserFS 4 with meta-data and hopefully FM’s to take advantage of it…this coupled with all of Linux’s inherent advantages.
I’ve been hearing about Linux having desktop potential and being ‘ready’ for the desktop far longer than I’ve been hearing about Longhorn missing ship dates and being pushed back.
Of course in time I feel that Linux may well be ready for everyones’ desktop. Just as in time I believe longhorn will be released.
I’ve personally enjoyed the long release cycle between Windows versions myself. Its nice to not upgrade an OS every year or two or have everything change drastically in a short period of time.
XP runs fast on current hardware, and when Longhorn is released we’ll see another push to upgrade all our hardware. I can live without that for another 18 months myself.
I’ve personally enjoyed the long release cycle between Windows versions myself. Its nice to not upgrade an OS every year or two or have everything change drastically in a short period of time.
—
you realise that you decide when you upgrade and that you can feel free to skip releases if its not required. ?
You have to look at things in context. Active Directory as a feature might have been ambitious (I wouldn’t know, I’m not a server admin), but it was the single major feature in Windows 2000 that hadn’t appeared in a previous Winodws release. For something that was four years in the making, do you really think Active Directory + ported Windows 98 features is enough to call the entire release “ambitious?” For comparison, the time between OS X 10.0 and OS X 10.4 will have been about 4 years. Taking into account the massive improvements between 10.0 and 10.4, can you really still say that Windows 2000 was an ambitious release?
Because I don’t see how one can switch from 2D ortho mode to a 3D projection matrix.
It’s not in the Cairo API, but the ability to do so is a fundemental property of OpenGL. If Longhorn ends up supporting it, and it turns out that it is an interesting feature to have, I don’t see any reason why it couldn’t be added quite easily. In Qt 4.0, the functionality is already there, because the API is designed to let you use QPainter and gl* calls together.
Afaik, in ortho there is no depth. So switching to 3D would leave with a 2d plane of windows; it would look silly.
Well, that’s why I said “cheezy” special effects. For example, to implement Expose, you could switch to perspective mode, put all window buffers sufficiently far away to make them small enough to fit everything in, and then have a cool “fly towards the screen” effect when a particular one is selected. Afterwards, you’d of course switch back to an orthographic projection. Regardless of how useful it is, the point is that if Longhorn can do it, there is no reason Linux cannot.
Also, someone mentioned that CPU to AGP is 1GB/s but the GPU can handle 24GB/s. How will you ever use that 24GB/s if you can’t send data to the GPU Fast enough?
You don’t send raw pixel data constantly to the GPU. You send a smaller, higher-level description. For example, say you want a gradient-filled button that is 100×50 pixels. At 32-bit color, that is 20kb of data. However, even if nothing was already loaded into video memory, it’d only be a few dozen bytes (not including overhead) of vertex position and color information that you’d have to send over AGP. Another example: say you want to composite an 1000×1000 image onto another 1000×1000 image. This operation chews up 12MB of bandwidth, which means that a GeForce 6800, which has 35GB/sec of bandwidth, can do it close to 3000 times per second. If all your images are already on the graphics card (which they often will be — the 6800 has 256MB of memory), then the only data that needs to go over the AGP bus is perhaps a few hundred bytes of vertex and control data.
The GPU was not build for things like 2D desktops.
Let’s be clear. The GPU was built to acccelerate transform and rasterization of triangle and quad meshes. If you can specify your graphics problems in terms of triangle and quad meshes, you can speed things up greatly using the GPU. Fortunately, it has been shown that 2D, specifically a 2D imaging model ala SVG or PDF, can be specified quite well in terms of triangle and quad meshes, by tesselating 2D shapes into meshes, and depending on the hardware to accelerate alpha blending and transparency.
In addition to the Glitz paper, take a look at the SVGL paper, which is as far as I can tell one of the pioneering works in this area:
http://www.lri.fr/~fekete/ps/svgl.pdf
I doubt it’ll even be more advanced than Quartz Extreme
It’ll definitely be more advanced than Quartz Extreme. The latter does all drawing in software, and only uses GL for compositing. Longhorn will do all drawing in hardware.
You wrote:
“Longhorn will do all drawing in hardware.”
You must be omniscient or something. AFAIK, Longhorn’s feature list is quite up in the air (if confused, read the article that spawned this thread)
BTW, you’re either a cocky, know-it-all twit, or someone actually working with windowing technology… which is it?
No omniscience is necessary here. The whole point of Avalon is to do drawing in hardware. Read Microsoft’s WinHEC presentation:
http://itlo.uq.edu.au/docs/TW04006_WINHEC2004.ppt
Taking into account the massive improvements between 10.0 and 10.4, can you really still say that Windows 2000 was an ambitious release?
No, is the simple answer. 2000 was simply what NT 4.0 should have been to start off with, and AD was simply an answer to NDS that Microsoft felt they needed.
However, AD has never been really used in any meaningful context. The vast majority of networks are simply domains with AD as a back-end storage area (that people don’t know about anyway), but people don’t use the features it has (which isn’t many). The people who knew they needed directory services used Novell.
“Give us examples. What sort of apps do you use, and which programs do you consider to be best in class?”
CAD is a blank area on your FOSSy-world-map
till now i’ve used:
Unigrafics
Pro-Engineer
Catia
Solid Edge
Solid Works
Microstation
AutoCAD
Inventor
You have to understand how Microsoft thinks. They see upcoming threats — and they deal with them head-on. Their solutions are often half-baked and lacking … but the PR that they generate often blunt any attack from a competitor.
“Directory Services” was the buzz around 1999-2000. MS stopped any possible incursion of Novell into its server business by introducing AD. It doesn’t really matter whether that business evolved much. Nobody really cares that much about AD or NDS anymore. Novell isn’t really a threat to MS anymore (don’t even talk about Novell’s Linux business — it simply isn’t mature at this point).
piracy for all products of Micro$oft
I hate Microsoft as much as anybody, but isn’t getting people to pirate their products going to lock people in just as much as if they paid for them?
But CAD isn’t a blank area on Mac.
ArchiCAD, VectorWorks, Sketchup and many other programs which you may find some references to at http://www.architosh.com/
Not that you have reason to change platforms. Once we have taken the time to learn one set of tools, it’s hard to justify starting over.
SketchUp though, has received enough praise that you may find it worth checking out:
http://www.sketchup.com/
No CAD programs for Linux? I beg to differ:
http://www.tech-edv.co.at/lunix/CADlinks.html
you realise that you decide when you upgrade and that you can feel free to skip releases if its not required. ?
I write windows apps for a living, so I tend to stay up to date with the latest windows version.
I’ve enjoyed the long life cycle of WinXP, nothing drastic has changed and I’ve spent a great deal of time just writing better code and debugging what I’ve already written then trying to get a handle on new OS features that my applications will need to support at some point.
To each his own. Some people love short runs between OS releases. I’m not one of those people.
“No CAD programs for Linux? I beg to differ”
None that absolutely don’t suck ass……
i think you misunderstood me.
i didn’t say there are no CAD-tools availabel on linux/mac/whatever. e.g. Pro-E has a linux-edition.
my point is that there are no free and/or open source ones.
those wich are availabel dont go beyond autosketch-level.
“Directory Services” was the buzz around 1999-2000. MS stopped any possible incursion of Novell into its server business by introducing AD.
Microsoft were afraid of incursions into their client business, not their server one. Lots of businesses running Windows with NDS/Zenworks clients without a viable Microsoft answer, and with the possibility that migration from Windows would be easier, was just not something Microsoft was prepared to contemplate. It would have pretty much stopped their server growth as well as it wouldn’t have dictated Windows, but Linux has already done that.
Despite how Microsoft hums and hars, and all those Gartner reports listing market share by revenue, Microsoft’s server business isn’t all that impressive.
None that absolutely don’t suck ass……
Why, have you tried them all?
my point is that there are no free and/or open source ones. those wich are availabel dont go beyond autosketch-level.
There are a few good free ones around, but so what if they aren’t top notch? If you want a good CAD package for professional use then you pay for one. You’re not making a very good point.
“Despite how Microsoft hums and hars, and all those Gartner reports listing market share by revenue, Microsoft’s server business isn’t all that impressive.”
What do you consider “impressive”….???
They make more cash on their shit than ALL of the Linux and Unix server systems combined.. We don’t need Gartner to tell us that….
i didn’t say there are no CAD-tools availabel on linux/mac/whatever. e.g. Pro-E has a linux-edition.
my point is that there are no free and/or open source ones.
those wich are availabel dont go beyond autosketch-level
The original point of this particular thread was begun by Darius when he said:
Hell, give me superior apps, and I’ll run whatever opearing system you want.
So where the apps are available on more that one platform, it would make more sense to use the alternative to Windows, so as not to be inflicted by spyware and viruses. More uptime, less hassle, and greater productivity.
Sorry, my mistake – you did in fact mention FOSS, not Linux.
my point is that there are no free and/or open source ones.
those wich are availabel dont go beyond autosketch-level.
Have you tried the latest version of qcad? It’s pretty good, even though it’s no AutoCAD…It’s certaily better than most the CAD free/shareware you’ll get for Windows!!
As another poster said, if you need professionnal quality software and there’s no FOSS alternative, then you fork out the bucks and buy a commercial one. (On the other, hand, the FOSS alternatives will continue to improve the more people use them, so you never know how it will be in a couple of years for Linux/FOSS CAD.
http://www.microsoft.com/presspass/press/2004/Aug04/08-27Target2006…
Microsoft Announces 2006 Target Date for Broad Availability Of Windows “Longhorn” Client Operating System
Windows WinFX Developer Technologies Will Be Made Available For Windows XP and Windows Server 2003
At a meeting today with several hundred of the company’s top developer evangelists from around the world, Microsoft announced that the Windows WinFX developer technologies, including the new presentation subsystem code-named “Avalon” and the new communication subsystem code-named Indigo, will be made available for Microsoft(R) Windows XP and Windows Server 2003 in 2006. This availability will expand the scope of opportunity for developers by enabling them to write applications that can run on hundreds of millions of PCs, resulting in enhanced experiences for users of those operating systems.