Has the Development of Window Managers Slowed?

al3x asks: "When I first got into Linux nearly five years ago, the new releases of competing window managers (like Blackbox, Enlightenment, Sawfish, etc.) were a constant thrill, and great strides were made with every release. I can't count the number of nights spent trying to get that sexy new E build to work, and what fun it was! But these days, window manager development seems to be stagnating. The last stable release of Enlightenment is from last year. Sawfish hasn't done much of anything in months, nor has Blackbox. WindowMaker had a recent update, but not with any exciting new features (it is rock solid, however). Now, verging from the paths of window manager favoritism or "they haven't been updated because they just work," why has development in this arena slowed to a crawl, and what's on the horizon?"

Comments

Enough already?

[Phil Wilkins]

Could it be possible that yet-another-window-manager, just isn't a particularly interesting project to work on any more?

We've had the pre-cambrian explosion, time for the mass extinction.

Golem

[frohike]

One of the promising window managers that's (IMHO) up and coming is Golem, being developed by a friend of mine. It's very simple but all of its features are provided by fast plugins. This is kinda like Sawfish, but without the overhead of a Lisp interpreter. Anyone looking for a new WindowManager to try out (and develop on) ought to check it out!

Couple possabilities

[nuintari]

I can't speak for all fo them, but AfterStep is undergoing a pretty big change, and as far as I know Enlightenment is getting a total rewrite. That'll slow down developement something fierce. Then ya look at some of the minimal ones, notably Blackbox and Sawfish, they both do what they were intended to do. new features aren't in the focus of some of the more minimalistic projects, so anything at this point is bug fixes. I don't know much about WindowMaker, but they could be working on a very new release, which could be in pre alpha states right now. Check the CVS of your favorites out and take a look, some of the code in the afterstep 1.9 is just great, but last time I checked, I still couldn't get it to compile completely.

Re:Couple possabilities

[Glytch]

Windowmaker is pretty much in the "add little features, fix little bugs" category from what I've seen in the changelogs. 0.70 was just released a few days ago, and (just speaking anecdotally) it works flawlessly. I never even have to touch the config files by hand anymore, the WPrefs tool works fine. And coming from a die-hard Slackware control-freak like me, that's saying something.

I think the development team should just declare it to be version 1.0. Windowmaker is stable and full-featured enough for it. :)

Re:Couple possabilities

[Brainchild]

I never even have to touch the config files by hand anymore, the WPrefs tool works fine. And coming from a die-hard Slackware control-freak like me, that's saying something.

And if you're even more of a control freak than that (like i sometimes am), you can use the wdwrite utility to store selected preferences, and then use the WPrefs tool and whatnot to handle the rest.

I think the development team should just declare it to be version 1.0. Windowmaker is stable and full-featured enough for it. :)

Close, but not quite. Alfredo's working on support for the _NET_WM spec (or whatever the devil that Grand Unified Post-ICCCCCM Extension is called nowadays), to support next-generation KDE and GNOME and whatnot. Once that's there, together with up-to-date documentation, then i would agree with the v1.0 thing.

Re:Couple possabilities

[ameoba]

[blockquote]
Then ya look at some of the minimal ones, notably Blackbox and Sawfish, they both do what they were intended to do. new features aren't in the focus of some of the more minimalistic projects, so anything at this point is bug fixes.
[/blockquote]
Which brings up one of the big differences between Free software and Commercial software; Since there is no real revenue stream to maintain, there isn't a long string of marketing dictatated releases after the project reaches maturity. TeX is a notable example, freezing the features, and working towards bugs (and extending the version number towards Pi).

I'm sure everyone here has heard the "release early, release often" mantra/slogan which applies to the early part of the development process, but what universal wisdom about the back-side of the dev. process do we have? We can follow the steps of TeX (as it appears many window managers have), and be happy with a solid, stable program, or we can take the path of feature-bloat, and keep adding things because they're nifty. (Mozilla comes to mind..)

They matured

[C0vardeAn0nim0]

KDE today is as good as window manager as MS windows or Apple finder, Gnome is getting closer, WindowMaker is rock solid and is small and fast, as Kojima dreamt years ago.

Until someone comes with a unbelievable great idea, things will go slow for a while.

And since the window managers "market" (don't know if this word can be applied to open source) are stable now, only the best and most used WM (gnome, WindowMaker, KDE) sees any development.

GO KDE!

[ekrout]

I think that this is primarily because desktop environments have taken over. The KDE folks must be hooked-up to caffeine IVs based on how fast they release updates and totally new applications. Ximian GNOME is also quite nice, and has a large following. Basically, "window managers" as we know them have been replaced by these more full-featured environments that are helping to bring Linux to the desktop.

Re:GO KDE!

[Nelson]

That's correct. In fact, I'd go as far as to say the the very concept of window managers has been a hurdle slowing down the progress of the GNOME.

For a while there was a debate about "the GNOME window manager" and then there was the whole E thing when people were getting frustrated and quitting their jobs. Sawfish came out to fill a void and that's what it has done and now it doesn't seem as important that it get's all the newest wiz-bang gadgetry.

Newer versions of E are sounding more and more like they are trying to build a desktop or new environment than simply a window manager, that's great, more power to them.

But the essential truth is that for the majority of Linux users the window manager concept has come and gone and that desktops are where it is all happening now.

e17

[technomancerX]

Seems like following the progress of e17 is interesting enough in and of itself... something about a multi-threaded open gl accelerated window manager what also includes a file manager and full widget library strikes me as 'making progress'. Not to mention the coolness of the underlying libraries...

.technomancer

Why E has slowed

[Raster Burn]

I've been lurking on their devel mailing list (check out their project page) and E's progress has slowed because of the recent downturn the tech industry. Open source programmers need jobs too....

Re:Why E has slowed

[cymen]

They just loose hope, fire up Windows and load PowerPoint to start a job presentation, and then get a deadly macro virus that kills their hard disk with the dreaded click of death hex code patch...

Seriously - whenever there is a problem just look to Microsoft for the cause.

[note: for the human impaired this is a joke, of course it may or may not be funny]

IceWM?

[X-Dopple]

I've been following IceWM rather closely for the past year, and it's at 1.0.9 - released a week or so ago, IIRC.

Really, though, what features are there to add to window managers? If you add too many features, then you end up like Enlightenment, which IMO is more like a desktop environment than a window manager.

Fvwm2

[gweihir]

I don't know what the problem is. I am still pretty satisfied with fvwm2. And some features you don't get in. e.g. KDE. Like the virtual desktop (FvwmPager) with the individual desktops actually being next to each other. Great for very large windows! And the switching mechanism (hit the border with the mouse) rocks!

I don't need all these graphical, slow and unintuitive menues. I am completely satisfied if I can add the shortcuts I need in 5 minutes to the pop up menues and have all the desktop space for my own use.

And I don't want to redo customization all the time. Basically I have had the same Fvwm2 configuration for years, with only small modifications. That means I can find everything very fast, because I know where things are!

Re:Fvwm2

[Rich]

You might like to try using the NET enhanced version of FVWM. This adds supports for the extended WM hints specification agreed to by both KDE and GNOME. The homepage is http://fvwm-ewmh.sourceforge.net/. The existence of this extension also shows that while WM development might have slowed, it hasn't stopped.

XFce development is still proceeding

[Stinking Pig]

XFce.org -- mostly incremental improvements, as you'd expect from an aim of small and fast, but recently anti-aliased display is supported and a migration to the ROX filer is about to be completed.

Because nobody's willing two do two things.

[Bowie J. Poag]

Here's why the mainstays for Linux development have ground to a halt:

1) Nobody is willing to work on something, pouring hours upon hours of work into it, only to have someone working in Company X take their code, and make a living off of tweaking it. Suppose you're writing a windowmanager for Linux. In order for your windowmanager to succeed, it probably has to be GPL in order for it to really catch on. And if its GPL, surprise-surprise, there are employees of parasitic companies like VA Linux Systems who make a nice living playing with your code. No one in their right mind is going to do something for free, working side by side next to someone who is getting paid to do the same. By simple virtue of the fact that parasitic GPL companies exist, you're effectively letting someone else make the money off your work by making it GPL. This is why companies who capitalize on Linux software development are a (tm) Bad Thing, because they assert a choking influence over the entire community. It stops becoming an exercise in fun, and rapidly becomes an exercise in profiteering.

2) Nobody is willing to think about doing anything different, more useful, or more ergonomic right now. The main driving force driving Linux UI development is "lets make it look like Windows!" which is a horrendously bad move. Instead of giving Linux its own face, its own appeal, and its own distinct look, we're playing Poor-Man's Explorer with X11. Instead of putting our own talents to work, making something useful for us, we're playing second fiddle to a third rate design by copying it.

Now, rather than purely bitching, here's what you can do about it:

Start at the ground up. Get ahold of the source of a weak windowmanager like fvwm, that has all the basic guts you need to work from. Ask yourself what makes sense to you as a user, NOT what makes sense because you've seen the same thing in Windows. Give Linux its own look. Try to avoid imitating other platforms. Build it because it makes sense to build, not because "Windows has it". The sheer number of things that Windows has wrong with its UI would require a completely separate article to discuss them in detail. Think about how to represent things differently. Is there a better way to represent the same information? Do you really want an OS that resembles a browser? Think, ask, and move. Learn, modify, and repeat.

Cheers, (and yes, Propaganda is still running..)

Re:Because nobody's willing two do two things.

[BenHmm]

1) Nobody is willing to work on something, pouring hours upon hours of work into it, only to have someone working in Company X take their code, and make a living off of tweaking it. Suppose you're writing a windowmanager for Linux. In order for your windowmanager to succeed, it probably has to be GPL in order for it to really catch on. And if its GPL, surprise-surprise, there are employees of parasitic companies like VA Linux Systems who make a nice living playing with your code. No one in their right mind is going to do something for free, working side by side next to someone who is getting paid to do the same. By simple virtue of the fact that parasitic GPL companies exist, you're effectively letting someone else make the money off your work by making it GPL. This is why companies who capitalize on Linux software development are a (tm) Bad Thing, because they assert a choking influence over the entire community. It stops becoming an exercise in fun, and rapidly becomes an exercise in profiteering.

Why do they start in the first place, then?

If a developer of a GPL project stops working on it, because a co-developer is in the lucky position of being paid to work on it, or because a company takes their great code and incorporates it into the product they need to sell to stay in business, then why did they start working in Open Source to begin with?

I'm not being stroppy, I just don't understand the psychology.

Re:Because nobody's willing two do two things.

[Quarters]

I hear over and over again this call to arms, "Design a desktop that makes sense. Don't just design one that looks and runs like Windows."

Well, what would that be, exactly? There are a few basic tennants of a GUI that appear similar on all platforms:
Text boxes
Combo boxes
Drop Down menus
Radio Buttons
Check Boxes
Scroll Bars
Buttons
Tab Panels
Icons
Shortcuts/Aliases
Start Menu/Apple Menu/KDE Menu/GNOME Menu
etc...

These are basic items that are the foundation of a GUI. Yet, when people implement these things we get the cries of, "That's just a poor man's Windows. Create what the users *want*!"

Well, what do the users want? Don't you think that Apple and Microsoft have invested quite large sums of money figuring out what the users want? Realistically, in this day and age, if you build a GUI that completely changes the paradigm of a desktop with items on it, folders, widgets, etc... you better have an idea that immediately resonates with everyone. Else it will look alien and nobody will use it.

So, why doesn't anyone ever list the items that would make a desktop that would be Linux's own? Isn't it about time for somebody to pony up with this grand vision, instead of just crowing about the fact that we should all be reaching for this mythic concept?

Really, what is it that we should be doing? Which path should we be taking to achieve this epiphany in UIs?

Re:Because nobody's willing two do two things.

[Bowie J. Poag]

I tried. Twice. Both efforts failed, largely because people can't seem to look past Windows as the one and only example of how a UI should be done.

For the record, the last "real" desktop I ever used was AmigaDOS 3.1. Fast, elegant, simple, all-encompassing, good design, clearly understandable, flexible, extensible and neat. The closest thing i've been able to look like it is WindowMaker, and even WindowMaker doesn't quite have it right.

A windowmanager need not occupy anything more than a single slat at the top of every screen. Why the top? Simple. The human eye, in Western cultures, tracks diagonally from northwest to southeast whenever it encounters an image. The flow of information should conform to that--Its absolutely opposite in Windows, where the origin of an action begins in the southwest corner (the Start button) and traverses awkwardly northeast. By the way, dont whine about "Well, what about non-Western cultures??? Are we just going to leave them out???" because the answer is YES. Let them come up with their own design. We do it our way, they do it their way.

A book is a perfect example of a proper user interface that has undergone hundreds of years of refinement. The title is at the top, relevant information is in the corners, and the page (or screen, if you will) is dominated by the body of the data. UIs should follow this convention.

Suppose you want to do a simple action. Start a program. In Windows, there are no less than 7 or so ways to start a program. Sometimes its an icon. Sometimes in an icon in the Tray. Sometimes its an icon in the Quick Launch bar. Sometimes its in the task bar. Sometimes its in Explorer. Sometimes its in the Start Menu. Sometimes its in DOS. On, and on, and on, ad infinitum, ad stupiditum.

A computer's UI should look and react like a television set, where all the channels are nothing but top-down views of books. Each channel has a single line across the top. It shows memory usage on the left, a date-clock on the right, and a single [x] button to kill the whole fucking thing and drop down to console. The remaining 99% of the screen can be occupied with any number of windows. No Docks. No taskbars. no trays. No icons.

All programs that exist on the system can be listed in a single pull-down menu. Right-clicking anywhere on the backdrop of that "channel" (or workspace) will give you the option of selecting a program to launch from a menu. A single, authoritative way of launching a program, not 7 of them.

Suppose you want to delete some junk--Fine. You need a filemanager. Not a filemanager, a browser, a text editor, a Trashcan, and a "delete" command. The filemanager is listed no differently than any other program in the menu listed above. One way for all. If you dont like it, use another OS.

Those are just two simple little improvements that would simplify the task of using Linux with a GUI a hundredfold. More options don't always means more flexibility. More options ALWAYS mean more complexity, and more intimidation for first-time users.

What I basically described to you is AmigaDOS 3.1's appearance in a nutshell. Installation of new apps was a snap, and it all worked out of the box. Instead, Linux has two maddenly different standards that fight for the same square foot of turf and both look retarded in the process. Until that gets resolved, you and I are stuck.

Hopefully

[tjansen]

Hopefully.. there's only so much you can do with a window manager, and 50 different should be enough for everybody, so people can now work on improving desktop environments or applications.

I disagree

[Greyfox]

Very few of us have monetary motivations. If you think an OSS developer has monetary motivations, you really don't get it. If you're an OSS developer with monetary motivations, you really don't get it either.

The reasons, I think, are twofold:

1) It's been done. You can find a window manager out there now that can do just about anything. There's not a lot of "Interesting" problem domain left.

2) There is no itch. My current window manager suits me fine. It does everything I want it to do. I don't really see the point of starting from scratch to code a new one. If I were going to fork a window manager, I'd start with the one that was closest to doing what I needed done.

Re:Configuring Window Managers

[MintSlice]

I recall installing drivers for my ati rage 128 a few months back:

Windows 98: 10 minutes
Xfree86: 4 full evenings

I recall my recent attempts to install my new Nvidia Gforce MX 200

Linux: No reboots. The new card (the old one was a Voodoo 3) was detected during the boot sequence after adding the card. I configured the card for my setup and it started X using the new card without missing a beat. Time taken: 30 seconds (literally)

Windows: 4 reboots, two failed shut downs, a trip into Linux to download drivers because the ones on the disk didn't work, and finally it worked. Time taken, about an hour (including download time)

Granted, I then had to install the Nvidia version of the liunux drivers to get full value from the card, but that was relatively simple and didn't involve a reboot of the system to achieve, and took less than 30 minutes including downloading the drivers. This process wouldn't even be an issue if Nvidia would GPL their drivers so they could be included in the kernel and X. Yes, that's right, the only hick-up I had with installing a new graphics card under linux was problems caused by non-GPL software.

Here's another example

A recent change of network card on:

Linux: restarted computer, the change of network cards was detected during boot, the current network config was migrated to the new card (after asking me if I wanted too) and the connection to the network was made without a reboot. Time taken: 10 seconds

Windows 98: The new card was detected, a driver was requested (it was on a floppy) and the machine finished it's boot process. No network connection. I open the network config tools to find that a secondl network card config has been loaded (which doesn't work). I have to remove the original setup, config the new setup and reboot. Time taken: 10 minutes.

Still not convinced? Try it with a sound card.

Sadly, I think Linux gets a hard rap when it comes to hardware. Changing and installing hardware in Linux is exceptionally easy, and limited only by a lack of support by hardware manufacturers. Given support by manufacturers linux ability to hand new hardware or hardware changes leaves Windows for dead. I can even change my processor and motherboard anmd linux will get everything sorted out during the restart without having to reboot once. Try that in Windows!

Oh and before you start telling me that Linux doesn't support as much hardware as Windows, try installing windows on a Macintosh, or a mainframe, or a wristwatch. Many operating system companies won't even move to a new platform for fear of what new hard might do to their stability. Apple's reliability on Macintosh is often cited as a result of a limited hardware base (an that reliability isn't that good. Microsoft originally started NT on an Alpha and said they wanted a kernel that would be easy to port to different processors. They don't even support the Alpha chipset anymore and are having troubles getting Windows to run on x86.

We've got Linux running on three different platforms here at work (x86, PowerPC, M68K) and running well.

What?

[Tom7]

I think this is a troll, but I think I should at least be a voice of dissent...

Lots of people, including me, work on software or do research for free, and don't mind when companies profit from our code or ideas. Mainly, this is because we believe that there is a great deal more work necessary to turn code or research into a product, and that work is primarily very tedious. I like the idea of a company using my code (I don't know of any who do, but I would) because they do work that otherwise wouldn't get done.

Second, I actually think the Windows UI is pretty good. More importantly, it is standard, which means that I can use KDE without reading any documentation. Regardless of how it might revolutionize the world (I don't think it would; the UI is pretty superficial and pretty subjective), new users are not going to switch to linux if they have to learn a lot just to use the UI.

So, I'm not saying that your opinion is wrong, but that asserting it as an "obvious" truth is.

Define weak.

[ErfC]

I use FVWM2 as my window manager of choice. It's fast. It gives me amazing and easily-customizeable control over everything. What's "weak" about it? What more do you need in a window manager? (I've never found myself saying "I wish FVWM could do foo...")

I'm not trying to disagree with you or berate you or flame you or anything. I'm honestly curious -- what's FVWM2 missing? What's wrong with it? What would you do to it?

Ubiquity breeds transparency

[Dag Maggot]

Asking if the development of Window Managers has slowed is like asking if the development of television remote controls has slowed down.

Window Managers have faded into the background as it is the tools and information inside the windows that (rightfully) recieves the focus. Since the advent of the Mac, the incredible uniqueness of windowing and the desktop metaphor in general has meant that we've spent an exorbinant amount of time focused on the UI itself instead of the tools contained by the UI.

To put it another way... imagine I was a caveman transported to today and placed inside of a room with a window. First, I would marvel at the incredible transparent substance that formed a barrier between me and the outside world, but after a while, I would take it for granted and simply use the window to see outside.

Xinerama Enhancements

[smartin]

One of the most promising new enhancments that I've seen are the Xinerama patches for WindowMaker. These allow for intelligent placement and management of windows on a dual headed setup. A Xinerama aware window manager will not pop up dialogs that cross between two monitors and will try to keep dialogs on the same screen as their parent.

I'm not sure if any of the other managers are working on this but it should be really cool when it is released.

Choice is good until you get it...

[mgkimsal2]

A bit late to the party, but these comments have 'gotten my goat' so to speak.

The majority of comments I hear from opensource/Linux people is 'choice is good', 'no choice is bad', 'I choose to do things in manner X', etc. However, having those choices in WM front ends apparently is NOT a good thing. Apparently we need just 'one' way of doing something (actually, I'm not all that opposed, if everyone would just write to that standard instead of bitching about it).

But... here we have an OS which is accused of being monolithic (Windows) yet it's also being criticized because there's more than one way to launch a program? So - if they lock you in to one method - it's bad. If you have choice - it's bad. Is this only because it's MS?

Re:Choice is good until you get it...

[_Sprocket_]

But... here we have an OS which is accused of being monolithic (Windows) yet it's also being criticized because there's more than one way to launch a program?

My first reaction was exactly that. But then it dawned on me that there's another way to look at this.

Say you're running Windows and need to launch Program X. There are 7 (I think that was the number used) different places you may have to look to run that app. Sometimes there's more than one way to launch the app - sometimes there's just the one. This means there's potential for a user to have to hunt around before they're able to figure out how to launch the app.

Having said that - I would agree that having only ONE way to do something isn't a plus. One standard way would be good. But giving power users additional ways would be my choice. I'd be a bit miffed if the only way to launch an app was through the Start button (or its equivilant).

Window Managers are mostly irrelevant

[nullity]

The glorified stature of the window manager is an odd architectural quirk of X-Windows. X-Windows is fundamentally concerned about, well, windows, so the window manager was pretty much *the* only application that every X system was running. Consquently if you wanted to add some feature (say a virtual desktop pager), you tried to get it into the Window Manager, because the window manager was already always running.

This "window manager is everything" view is actually sort of primitive. Most advanced operating systems have turned the window manager into a really mundane implementational detail that even programmers hardly care about. BeOS, Windows, MacOS, etc.

I hope this trend continues to GNOME & KDE...and we see the disappearance of those insanely bloated window manager preference dialogues and see the window manager behaving like the submissive quiet little technical detail it should be (at least from a user's perspective). Check out Havoc's latest project "Metacity" for an example of a well behaved CrackFree(TM) window manager.

-Seth (gnome usability project lead)

Re:Improve Xfree86

[valdis]

There's two major problems here:

1) It's fine to say "use a GeForce3" if you have one. However, the majority of us do NOT have a GeForce3. For some of us (like everybody who's not on an Intel-based system), a GeForce3 may not even be on option. Enlightenment 0.16.5 works fine on my AIX box - it even runs fine on my old home AIX box (a model 350 - a whole whopping 66mz Power chipset). Remember that Sun and SGI are both shipping Enlightenment now, and even in the Intel-based Linux world, I don't think it's safe to assume any more capability than 24-bit color. E17 shouldn't be assuming effects - it should be saying "use options if the X server (which may not be XFree86) says they are available". Last I checked, even XFree86 didn't have XRender extension support for all drivers yet.

2) There's a major distinction to be made between "looks nice" and "is productively useful". Yes, it's good if your window manager looks pretty (I'd like to thank Christian Kreibich for his Ganymede theme) but at least in *my* case, I get paid to get things done on the computer. As a result, I *want* my window manager to look pretty, but I *need* it to help me get things done quickly.
Now, some features are pretty easy to demonstrate why they'd be useful - for instance, it *would* be nice to have sane support for drop shadows. However, the reason for drop shadows isn't "because it's cool" - it's because it's additional visual information that helps you identify the edges of windows and the actual stacking order.

Now maybe there's a good HCI (human computer interaction) reason for supporting motion blur and bump mapping for the window manager. And if there is, I'll be happy to listen. But keep in mind that you're talking about *window manager* controlled screen real estate, not application windows. There seems to be a backlash by at least some users who feel window managers take too much screen space away (I'm one of these - my current theme has only 5 pixels on sides and bottom, and 25 or so in the titlebar). Not much you can *DO* with bump mapping or motion blur in a strip 5 pixels wide ;)

Just keep in mind that "eye candy" and "usability" are not always synonymous....

Next wave: windowmanagers for handhelds

[gnrfan]

Hey.. Now a whole new group of windowmanagers is needed for all those devices running linux. Mandrake co-author of recent versions of Enlightenment was running Blackbox on his IPaq but has coded a handheld-specific windowmanager he's calling ePaq (http://www.handhelds.org/z/wiki/ePAQ)