I remember reading a quote from a Linux enthusiast, dismissing the impact of Longhorn, stating that the features new in Longhorn essentially exists in Linux in its present form. XAML--oh, that's just XUL--and so on. WINFS--oh, that's just a database.
I feel that, if Longhorn does deliver in its promises, it will severely complicate the future of Linux and Java. If the whole Windows world becomes managed, won't Java performance suffer a bit with two runtimes, two libraries, two garbage collectors needed to run a Java application. Will Java based UIs appear rather dated compared to new Avalon applications, or will Sun be able to match the Longhorn UI look and experience as well as maintain the same UI consistency across multiple operating systems?
For a while, Linux had the advantage of being a more solidly built operating system than Windows 9X; I don't believe it still maintains an edge anymore with Windows XP and Mac OS X. Can an army of volunteers keep up and maintain the same discipline as a focused set of developers in Apple and Microsoft?
Avalon is set to have a far richer set of graphical capabilities and will make full use of the graphics processor; the idea, of course, is that the UI will become as rich or even richer than current video game titles. This advance is proceeding quickly through Microsoft's close collaborative work with hardware developers such as ATI. I think Mac OS X proved earlier that major advances can from come better integration between hardware and software. Linux does not have a good history with hardware support and integration, so it will probably fall behind.
Is there any activity by Linux developers towards a rich multimedia interface? I wonder if the decentralized nature of Linux might even hurt it in this next presentation wave. It's hard to see any advances in this area unless IBM and Sun (perhaps with its Looking Glass) comes along to help.
I am getting a sense of deja vu. When Microsoft announced its Internet plans back in 1996, I had difficulty figuring out how Netscape would fend off some challenges. IE, for instance, was available as a reusable ActiveX control--a feature that won over AOL as well as a number of ISVs. There was also ActiveX documents, the Inet API, HTML integration everywhere, although Active Desktop, however, never caught on and appears to have been abandoned in Longhorn. Microsoft had a multipronged strategy, coming in from all angles with IIS server, FrontPage, MSN, Office, VS, Explorer, and Windows API. Netscape seemed to focus on individual features, while Microsoft focused on linkages, understanding that the world cares more about how to connect different products together. While IE could be easily popped into any application, Netscape's browser was not embeddable for a long time. Well, sure enough, Netscape is no more.
I do think Linux will still be around, if, for any no other reason, than that it's free and open-source. I just don't feel that it will be in many desktops, since I feel that Longhorn will present a more compelling interface. (I'm not talking about the current alpha interface, which is essentially the same as XP; the neat stuff is still hidden from most of the developers at Microsoft, so as not to leak out.)
Linux GUIs will not look dated if Sun deliver on Project Looking Glass
Posted by: RichB | February 24, 2004 at 11:13 PM
Since when has Linux been a factor on the desktop? Now or.. uh.. ever? The CEO of Red Hat cedes the desktop to MS in official on the record quotes-- even he isn't that crazy.
UNIX has had how many years to come up with a usable interface? Still not there, and it never will be, because 1) it's simply not in the command line canon and 2) usability is at least two orders of magnitude harder than writing an OS (as crazy as that sounds, it's true). Nothing about Linux is going to change any of this.
Now, Apple might be a better contender for "friendlyizing" UNIX, but Linux? It is to laugh.
Posted by: Jeff Atwood | February 25, 2004 at 12:25 AM
I know of a startup, Lycoris, in Redmond that had been trying to provide a friendly face to Linux. They do have dreams of making Linux a desktop machine.
Posted by: Wesner Moise | February 25, 2004 at 12:32 AM
Linux seems destined for cheap jackknife servers and as a base for creating highly specialized OSes in specialized hardware.
Posted by: theCoach | February 25, 2004 at 06:57 AM
I do not see Longhorn as a threat to Linux in the server arena. Linux beats Windows hands-down for large-scale, low-cost deployment for servers. Windows licensing is too expensive and too restrictive. In addition, new developments like Mono and will enable most of the functionality of ASP.NET and enterprise SOA in .NET to run and interoperate on Linux servers. There is also a product (can't remember the name) that allows CIL to be hosted in a J2EE container.
You can argue TCO on the server side, but every discussion on TCO I have seen smells like it was bought off.
What marketshare do you think Longhorn will take away from Linux on the desktop? Linux on the desktop is not its strong point right now. But, who knows what's coming next?
Also, I would like to know in what way you believe the NT kernel superior to Linux.
Casey
Posted by: Casey Marshall | February 25, 2004 at 07:36 AM
Well, as far as LINUX on the desktop, there is one interesting factor I didn't touch on. Amortized across time, hardware is free. Six years ago it was impossible to build a complete PC for under $500 including monitor. Now it's not only possible, but common.
That has unforeseen implications in terms of MS licensing costs: the cost of the hardware is approaching zero, while the cost of the OS license is constant (monopoly influence, perhaps?). So it's a larger and larger percent of the overall cost.
On a $2k PC, $75 for an OS license seems reasonable, but on a $500 pc? $75 is almost 1/5th of the cost of the box! That COULD force some vendors to explore desktop OS alternatives, however painful they may be.
Posted by: Jeff Atwood | February 25, 2004 at 08:38 AM
To me the architecture formally-known-as Palladium is the killer app for the Windows OS. If all your content venders move to it, where does that leave Linux?
Posted by: Ken Brubaker | February 25, 2004 at 10:58 AM
I am not qualified to the remark on the superiority of Linux's kernel over XP or vice versa, but I don't think there are that many meaningful differences anymore.
As for the marketshare, I can see Linux surviving in quite a number of desktops, actually, particularly those used for a limited business purpose, where client richness is not a factor.
Posted by: Wesner Moise | February 25, 2004 at 01:26 PM
There is some self-awareness about Linuxs UI shortcomings. Eric Raymond recently wrote an article addressing some of the problems, and solutions, he sees.
http://www.catb.org/~esr/writings/cups-horror.html
There are plenty of rich multimedia applications that use Linux as their base. Tivo is the biggest name that comes to mind. There are several Open Source PVR applications out there. MPlayer keeps getting better and better with every release (as one one hope it would) as a multimedia ripper/player.
There have been some recent announcements by various teams for X-window replacements. That's analogous to the Avalon re-work being done for Longhorn. X-Windows has always been much more transparent in terms of the client-server than the Windows display sub-system and it will be interesting to see where they go.
Posted by: Scott Koon | February 27, 2004 at 07:00 AM
Regarding WinFS being "just a database", I think whoever said that doesn't really understand what WinFS is all about. I'm not even sure *I* really understand what WinFS is all about, but I know it's more than just an extension to the alternate data streams that allow the metadata to be attached to files. I think it might be more comparable with the journaling filesystems that *nix systems have. Which is a good thing. I think the ReiserFS and, to some extent the ext3 filesystem, are much better than the NTFS system in terms of performance and maintinence. I've never had to "defrag" my Linux fileserver, but I have to defrag my Windows box every two or three months. :(
Posted by: Scott Koon | February 27, 2004 at 07:07 AM
http://www.catb.org/~esr/writings/cups-horror.html
Oh yeah, I saw a reference to that, assumed it was an average blogger, not "the" Eric Raymond. Interesting.
Anybody can build something that is difficult to use. That's no trick at all. But making it easy to use-- that's where the real work starts. It's hard. Really hard! Which is why usability is virtually nonexistent in *NIX; programmers are too busy working on the "fun" geeky parts and ignoring the difficult ones, eg, making it easy to use for an average person.
Tivo is a good example of how to make something easy to use, probably the single best consumer electronics experience I have ever had.
Posted by: Jeff Atwood | February 28, 2004 at 02:13 PM
I do wish Linux zealots would stop just bashing windows with words and actually get the product to a state where it is really competing with Windows. Just pointing out what a nasty companyM'soft is just aint gonna get the job done. Learn from M'soft, yes I said learn. "Know thine enemy" and whether or not its palatable; M'soft does do somethings ( a lot of things in fact ) very well. It's just the way they go about things that piss people off. So take your heads out of your ar** and produce something that "Joe Public" would wanna use. Half the battle is already won; Linux is free, robust and secure.
Posted by: P Sympson | March 13, 2004 at 02:31 PM
You gotta give the massively distributed net of code writers –geeks if you like- some credit. Linux has evolved and it will very much invade a good share of desktops really soon. A new Linux kernel is out every week. When was the last major windows kernel upgrade?
I have a recent laptop that I managed to get up and running using the latest production kernel 2.6.4. Every single peace of hardware that was shipped with my laptop is perfectly working; you name it, IrDA, Bluetooth, WiFi or Modem.
The sound I get is way better then what I get from Windows; thanks to the ALSA project. I can read/write from/to the slow non-defragged NTFS partition. Free software implementations have allowed me to run Audio over Bluetooth (OpenOBEX Project) –very difficult from Windows with my $200 3Com card- and play music by a click on a Sony TV Remote Control (LIRC Project).
Usability is there yet sometimes troublesome. Developers are progressively working towards perfecting the imperfect.
At work, a windows machine, the computing software we use runs on exceed. Duh!
Posted by: Muhammad Al-Ismail | March 29, 2004 at 03:45 AM