Slashdot Mirror
Run Gnome -- On Windows
Charles Northrup (followed by many others) writes: "After a long battle with the POSIX.DLL from AT&T's U/WIN, we finally have recompiled over 4 million lines of code for the Gnome Desktop running on Windows. We had to change just under 100 lines to do this. Amazingly, the recompilation of Gnome only took 1.5 weeks. Then we had to spend the next 3 months working on the 47,000 line POSIX.DLL to get Gnome working." Here are some screenshots of this odd hybrid, pointed to by reader dimator. (Northrup invites e-mailed comments and questions, too.) And jumbolo points out the discussion about this at The GNOME Project's Gnotices site.
Remember, this isn't a native port to the win32 platform. You'll need an X11 server on your Windows machine - and that's what does all the hard work.
What kind of video card do you have?
I used to have those issues. Then I bought a Voodoo 3.
I work at a Linux company. Nobody in engineering uses anything but linux. Nobody.
On the other hand, management, sales, marketing and the like use Windows. This is a Bad Thing(tm) because it looks bad for us -- customers come in and see our salespeople giving PowerPoint presentations on our Windows boxen about why we should support their (embedded) linux development.
We care a lot about migrating those managers to Linux. Right now we're using Win4Lin to provide that path -- but my point is, for us, the need is there.
It's a separate thing from this UWIN layer, but coding against straight GTK (without GNOME) and using the Win32 GTK port permits one to make Real Win32 Apps (tm). I've actually done a few that way -- only taken on average a few hours to port and work out the bugs. Not sure what it would take to combine that with this UWIN-based GNOME port.
In case it isn't a joke:
1.5 week to port Gnome to UWIN (i.e. make the necessary changes to Gnome to make it compile under UWIN).
3 month to port UWIN to Gnome (i.e. make the necessary changes to posix.dll to make it actually run the (newly compiled) Gnome).
Cygwin being GPL'ed and comming from one of large Gnome supporters would seem to be more "politically correct". Is UWIN techincally superior?
> If you come from NT, the graphics performance of
> X is scary.
When NT was nerd, a fellow nerd tried to show me how much faster it was than X. He took an NT application, moved the cursor back and forward over the menubar, and noticed how NT had no trouble drawing and removing the menus as fast as he could move the mouse. He then took an X11/Motif application, and moved the mouse over the menu bar. You could actually *see* the application draw the menus.
I was confused, because I had never seen X be that slow, but then again, I mostly uses xterm and Emacs. So I took Emacs, moved the mouse back and forward over the menubar (which I rarely use) and Emacs had no trouble at all drawing and removing the menus as fast as I could move the mouse.
The difference was that Emacs by default uses its own Xt based toolkit, rather than using Motif.
The morale is, despite being network transparant, X11 itself is not slow. However some of the toolkits are pure disasters.
Unless Gtk+ is as bad as Motif, it is unlikely that the performance alone will scare users away. The difference in Look&Feel are much more likely sources for nebie angst.
Of course, the issue is somewhat convoluted by the fact that X11 servers under Win often are much slower than "bare metal" X11 servers. I use an old version of Exceed, which is acceptable for text and GUI controls, but very slow for displaying images.
Check Cygnus for XFree for Win32. Apparently it only works on NT, ME, and 2K at the moment.
Stating on Slashdot that I like cheese since 1997.
Look, as soon as the CYGWIN folks release XFree 4.0 for Windows (which they're about to do), Windows folks will have a high quality free X server available. From there it's a simple jump to Gnome and KDE applications on Windows. That's the migration path.
A large IT shop can't afford to simply dump Windows and move their entire userbase to Linux in one fell swoop. There's training costs to consider, application testing for internal conformity within the organization, plus security and productivity issues. If you're managing several hundred to a thousand desktops you can't afford to segment off a department just to try out a new OS, nor can you afford to just move everyone in enmasse given the potential risks involved; what if it doesn't work? You need a migration path whereby both systems can be tested and either junked in case of failure. Successful IT shops plan for failure otherwise they experience failure.
Microsoft has done a good job at killing off potential migration paths away from Windows by either buying up competitors or squashing them with anti-competitive means. See: Netscape. This is a major threat to Windows because within a year or two it will allow an IT manager the possibility to deploy GNOME and its applications on the desktop and test them without being dependent on their success. Combine this with OpenOffice and Gnome-1.4/2.0 and the Free Software community will have provided a safe migration path off of Windows. Most IT organizations are risk averse, but they're not stupid about unnecessary expense. If they can get away from the Windows tax without affecting productivity, watch the world dump Windows.
Cheers,
J. Maynard Gelinas
If you teach people Gnome applications on Windows, you can then slip the OS out from under them without complaint.
Personally, I think the requirement to have an X server is terribly flawed... but then I hate this "desktop war" too. I bet they'll eventually be able to do it without relying on Exceed. Ethereal is a Gnome app and it works o.k. on Win32 without one. (http://www.ethereal.com/)
If you're using an X server why not just run X over the network?
It is so similar in capabilities to Xlib that I really am suprised that fewer toolkits are not ported. A large number of calls have 1:1 correspondence. The biggest problem is that their event model sucks, while Xlib's fonts suck, requiring a portable kit to use the worst of each.
Actually, most widget sets in X also create a separate window for each widget (button, menu, ...). I'm pretty sure Xt, Motif, GTK+, and Qt all use windows for most widgets. I recall that Xt/Motif called windowless widgets gadgets instead of widgets, but they were rarely used.
Software sucks. Open Source sucks less.
While running Gnome on Windows is a novelty -- you probably won't find a lot of apps that will make it worthwhile.
I know this is a technology demo, but you still have to ask yourself what could you do with such a setup? Most Windows X display servers do not support everything you'd get on Linux. It's been a while since I ran X on Windows, but a lot of apps had wierd repaints, and other subtle misbehaviours. While this is minor, other problems exist such as the desktop... you can't exactly discard the MS Toolbar (Start button) for the ole Gnome foot.
What WOULD be cool is if this could run without Windows... just run it on top of DOS, like Windows 3.1 did (and Windows 98 still does, though it's deliberately more difficult to run a non MS DOS 7 underneath the GUI).
Why run this on DOS when you could use Linux? DOS is simpler to install, runs on very old hardware (286's), and while lacking memory protection and multitasking... DOS was rather stable. Many older NT admins cut their teeth on DOS (course, the new ones think a system log is mouse turd behind the rackmount).
Anyone try this in Windows under VMWare? Heh.
Actually, this is probably more due to the priority you are giving your X server. I reniced my X server, window manager, and panel to -20 and got _really_ nice results.
Engineering and the Ultimate
after starting X, do
ps eafx
look for PIDs with the names "panel", "X", and "sawfish"
then, su to root, and for each PID, do
renice -20 PID
replacing PID with the process's actual PID.
Engineering and the Ultimate
It is indeed an impressive feat that they were able to get gnome to compile under the UWIN Posix library.
I suspect that Interix would have made the job much easier.
Interix (formerly OpenNT) was a posix-compliant subsystem for NT that provided many utilities and for all intensive purposes, a mostly complete unix subsystem for the NT platform. I managed to port WindowMaker and AfterStep, among other things, to run under Interix. Some of you may remember, but the people at softway systems once ran an add in a Linux magazine, announcing a 'bounty' system for people who could port certain things to their subsystem. Gnome was one of them, but I guess no one took up the challenge.
Unfortunately, Interix (Softway Systems) was bought by Microsoft last year some time last year. Microsoft said that they were going to roll the Interix package in with their "Unix Services for Windows", but I have not seen anything come of that. I often wonder if Interix was bought just to stop their progress?
I have some screenshots of the two window managers I managed to port, running in NT via Exceed:
http://billimek.com/interix/
I will never forget the ignorance I was met with when I went into efnet's #linux asking about porting a window manager to interix. All of the people in there ridiculed me and told me I was stupid. Some even said such clever things like 'man X' and other such nonsense. Thanks to their zealotry, I was determined to prove them wrong. And when I posted the first screenshot of a successful port for them to see, all I got was stunned silence.
First of all, the existing Windows interface is alright, you can virtualy do anything you can with Gnome, and it does look and run sleeker.
Bah. How about multiple desktops? Yes, I know there's a zillion of those I can find off nonags, and they're all buggier than a Bangkok summer. The one on the resource kit is actually the worst one yet. A DOS box with less than completely crappy terminal handling would be nice too (try running w3m or just emacs over win2k's telnet). I wouldn't mind being able to move windows with alt-drag also.
--
I've finally had it: until slashdot gets article moderation, I am not coming back.
In what sense ? MFC also adds a "layer of abstraction". Or maybe you're confusing GNOME with a Window manager. Repeat after me:
GNOME IS NOT A WINDOW MANAGER
The point is that if you want GNOME applications to run on Linux, you've got to port the GNOME development platform (yes, GNOME includes a development platform, it is NOT a Window manager)
How many of your professors run GNOME applications on their Windows desktop ? WOuldn't it be nice if they could do so, given that GNOME/KDE have all the momentum at the moment ?
No you haven't. You displayed it on the Windows box, but it was running on your Linux machine. The difference here is that the application is running on the Windows machine, so one can try out GNOME applications without needing to install Linux.
Uhm, that's a shell replacement, too.
-
Personally I think the native GTK+ port to win32 is cooler than the CYGWIN one. You don't need an X server. Granted GNOME would be a huge task to port since you don't have the POSIXness. Check out the Gimp and GTK+ ported to Win32.
This may be successful with people who are interested in WMs like Litestep, Darkstep and other less known ones.
Having tried most of them, I'd say that they're usually buggy and quite beta. Usability is limited, mostly because you lose most of Windows' native functions (which are dependent on explorer.exe, the WM) and only implement a limited number of new ones.
This feature implementation is usually done by using plugins and modules, which are in theory a good thing. In practice it means that you usually get a slow manager that can crash or behave badly just because of one buggy plugin.
I've managed to use Litestep for a few days, but eventually went back to explorer.exe because features like file managing and printing were still tied to the latter, making the whole idea somewhat flawed.
I hope Gnome works as one would expect. If it does, I'll probably change for good.
Flavio
Although I'm sure it'll be useful for someone, this strikes me as a really bad idea. One of the few things Windows has going for it is (gasp!) a common look and feel, something *NICes have never had. I can't picture any Windows users deciding they want to have a bunch of uglier widgets right next to ones that look totally different, just so they can run...um, gnumeric? Gaim? Gftp? What exactly would be someone's motivation to run this? Certainly Gimp would be ok, but in that situation the program itself should be ported.
I'm no fan of Windows, polluting it with the *NIX world's fragmentation seems counter-productive.
-lx
Cygwin has a large based of Linux/GNU utils and programs that have already been ported. I recompiled GTK+ and gimp for cygwin without having to change a single line of code - so cygwin is a relatively mature environment.
-josh
I think your mistake is assuming it's necessary to "win".
Choice of masters is not freedom.
There's a Java one:
:-)
WeirdX
It's monochrome, last I tried it, but it's better than nothing
Glückwünsche, haben Sie Slashdot ermordet, indem Sie zum korporativen Druck beugten und Subskriptionen einlei
--
--
Mod up a post Rob doesn't like and you'll never mod again
Windows folks will have a high quality free X server available. From there it's a simple jump to Gnome and KDE applications on Windows. That's the migration path.
Why would a Windows user want an X Server? Why would a Windows user want to migrate to Gnome? IT departments support Windows applications and no "migration" would realistically aim to dump those. I don't see the benefit for Windows users. They don't care about the GPL. They care about their hard work and data.. which happen to be in Windows apps.
cpeterso
The VALUE??? I saw the value in 1 second. I want to write some open source cross platform software... really, I do, this is not just a rhetorical example. Now that Gnome has been ported to Windows, I have a way to do my development under Linux with GTK and Gnome as the main target, and easily recompile my application to run under Windows-with-Gnome for users who can't or won't switch. I suppose that was already possible with QT, now that it's GPL, but for programmers who prefer the Gnome programming model, this can be very convenient.
My second problem with your post is that you think that there's too much focus on "cool" and "technical" projects. Well, most of the people working on that stuff are not working for some big company that tells them what to do. They like it that way, even though a lot of them are working for free. They probably won't be interested in working for free on what you want them to do.
A lot of Linux users don't care if we "win the war" against Microsoft. Why should they care?
Torrey Hoffman (Azog)
Torrey Hoffman (Azog)
"HTML needs a rant tag" - Alan Cox
One problem with this is that GNOME for Windows requires an X server presence on the Windows machine. Because of this, the port, while still pretty cool, loses some technical merit. Using an X server means that they had all the standard X libraries available on UNIX systems; they didn't need to port any low level Xlib stuff to Windows.
Because of this, I find that the Windows port of Gimp has more technical merit; it does not require an X server - it is a full-fledged port of the GTK library to the Windows platform.
Now I'm not a GNOME expert, but aren't most of the graphics aspects of GNOME built on top of GTK? Why not port GNOME to Windows using this available library? This way, an X server will not be required. (Of course, I'm assuming that non-graphical libraries, ORBit, glib, etc, that GNOME uses are easy to port to Windows).
> Why in the hell would I want to replace my superbly useful Windows Desktop with GNOME???
It's not just the desktop, it's the ability to run software written with GNOME. I'd imagine you can disable most of the visible GNOME stuff and just run the apps.
I'm in two minds as to whether this is a good thing or not.
Good: Makes it easier to write apps that work on Windows and Linux.
Bad: Makes it easier to write apps that work on Windows and Linux.
- Muggins the Mad
I thought that their claim of how easy it was to change the gnome source to get it to work was a sure sign of a hoax until I saw the list of requirements...
An X server...
:)
- Muggins the Mad
Actually, it says:
/. misquoting/representing and noone being able to check to see what really got said.
What makes this effort unique is that we did this in 1.5 weeks, and changed less than 100 lines of the original source code used in building GNOME for Linux.
Just another case of
cyn, free software and *nix operating systems enthusiast.
I don't get it, I've written a lot of programs using GTK+ and they recompile flawlessly on windoze using VC++. Occasionally I've had to do a #ifdef WIN32 here and there, but really it's about coding the thing portably in the first place. Hell, you can even get precompiled dll's for GTK from the web site! I guess GNOME has a lot of posix specific stuff in it. Whenever the requirements state that the program must run identically on windoze and *nix I always use GTK. Sometimes I'll do the actual development on the windoze box using VC++, sometimes I'll do it on linux, but I'll always make it a habit to do a compile on the other OS after I add each feature (every day or two). I mean really, that's the idea of the toolkit isn't it?
How we know is more important than what we know.
Interestingly, you can just make windows look like it's running gnome:
d erbig.gif
http://www.dtop.org/outsider99/img/sshots/gnomesi
________________________________________
Napster-to-go says "Fill and refill your compatible MP3 player", which is a lie. It's not MP3. It's WMA with DRM.
What I don't really get, is how come it takes 1.5 weeks to recompile à 4 million code program. Where I worked we had about 1 million or so line of code and the thing compiled in about 2 hours.
Consider..
I use gnome on X all the time. I also have a windows machine for those annoying apps that just refuse to be ported, and that I am not comfortable enough with the X-ish alternatives.
On my windows machine, I really dislike explorer. It crashes regularly, and doesnt react the way I would like.
This is directly why you can choose to replace your shell with another one. I tried with litestep, but it was somewhat buggier for the things I did (YMMV), and still wasnt 'what I was looking for'.
With this, I can have a more linux-oriented desktop, showing my truer colors and spreading the word, and at the same time, allowing me to have a more consistent feel from (windows) desktop to (linux) desktop.
I think it kicks BUTT!
GPL'd web-based tradewars themed space game
....or how about creating a standardized, free platform on which to build apps that runs on all unices as well as Windows?
Look at projects like Eazel, or Helix. If we have a standard gui and things like bonobo across all of the unices *and* MS machines, then we've just nailed pretty much the entire market. With the BSD core of the new MacOS, it's a hop skip and a jump from here to having Gnome capabilities on every major computing platform. I don't see the down side.
As for the car dealer analogy... If there's anything that the "technology age" is going to be remembered for, it going to be bad analogies. We're not talking about cars at all here, we're not talking about anything like them. Cars started out in a standardized atmosphere, and since then have become more and more and more proprietary.
We're talking about going in the opposite direction -- bringing a common, free shell/development environment to, once again, *all* of the major computing platforms.
I think that all the people asking about why this is cool should have their slashdot logins revoked.
This is the coolest thing I have heard in a long time...
1.5 weeks!!!!
It's mind boggling. Charles Northrup will go down in my books as one of the great hackers of our time.
Just the sheer audacity to look at those bazillions of lines of code and say, "Dang it. Today I'm going to port those things to windows." A lesser man would have died on the spot at the prospect of such a tremendous undertaking. I would have died if someone had even shown me that much code, let alone told me to port it.
Mr Northrup, you are one crazy dude...
OK folks. While the end result of this project is "cool" I fail to see the "value" that the end user will get from it.
I mean, do you think the Windows user is willing to learn yet-another-UI that sits next to the Windows UI?!!! Yes the technical group will be excited about this (as I am) but we make up only the very small % of the users at large.
Folks, call me crazy, but as long as Linux and the Open Source community keep there focus ENTIRELY on "cool" and technical-only-oriented projects, we will never be able to convince the Windows user to switch nor are we going to win new users.
Beside, the war is no longer againts MS's OS, it is now with MS's application as well as AOL's content offering.
-- George
Karma stuck at 50? Add 2-5 inches.. err.. 2-5x Karmas Count to your pen1es.. err.. Karma all naturally and private
They needed an X server and a commercial unix emulation package to do this.
KDE too started out relying on proprietary software but (with Trolltech's help) phased it out. Once XFree86 4 begins to work around buggy 16-bit code in Windows 9x (it already works fine on NT/2K) or Whistler is released, we will have our free X11 server for Windows. And it shouldn't be too much work to move from Uwin to Red Hat Cygwin, another POSIX layer for Windows.
Will I retire or break 10K?
XFree86 4 already works on Windows NT and Windows 2000; however, there is buggy 16-bit legacy code in Windows 9x's kernel and GDI that keeps the X server from working properly (according to the XFree86 for Windows FAQ). This should be less of a problem once Windows 2000 Personal (codename Whistler) is released.
Will I retire or break 10K?
The Windows equivalent of Galeon (i.e. a stripped-down web browser using the Gecko engine from Mozilla) is k-meleon (no, despite the K, it has nothing to do with KDE).
Will I retire or break 10K?
I noticed that in all the screenshots it reads 12:00 a.m. or 0:00:00 anywhere that time is displayed. Window names also appear to be replaced with 5 asterisks. Doesn't really look polished yet.
I also find it funny that they had issues with GIMP. I have no problem running GIMP native on my windows machine.
WARNING: there is a trojan on your
They needed an X server and a commercial unix emulation package to do this.
What exactly was the point? Prove that Gnome is not portable?
Let's clear this up right good.
;) as the window manager for now and, when the next upgrade of Windows comes out, I can slip in Linux on the backend and no one outside my skull would know the difference. World domination, one company at a time?
The article says it took a 1.5 weeks to:
a) Make the code Windows-playable
b) Compile the code
They had to hunt throught 4 MILLION LINES OF CODE to locate certain disparities between Microsoft's idio(t)syncracies. They changed only 100 lines, which is pretty good.
Now, yes, you do need another package for it to properly work.
Big deal.
I can use Gnome (though I am a KDE man
I used to be someone else. Now I'm someone better.
Real life is underrated.
"It is a greater offense to steal men's labor, than their clothes"
Ok, that's a flame of some sort. But don't you guys think that is kind of pointless? Why support Windows?? Why bother?? Linux stuff running on Windows.. oh my God! I can't imagine KDE doing the same thing.
First of all, the existing Windows interface is alright, you can virtualy do anything you can with Gnome, and it does look and run sleeker.
Second, why bother with porting code to Windows platform? What is the intention of this? Who are the possible users? Are those Linux zealots that have to work on Windows (NO WAY!!!) and customize native applications to support Linux shortcuts/commands?
Now moderate me down.
http://dtum.livejournal.com
er..no.