LevelsOfDetail

Sorry for the cliche heading.

Reflecting on my own situation gives me a good insight into why Linux hasn't really landed on the desktop. It comes down to one simple reason: If I run Linux, I can't pay anyone to solve my problems for me.

Actually, at work, Linux is great. At work, I have to solve problems that are new. Problems that nobody else has solved yet. At work, I need the best tools that are available. I need the ability to customize my workflow and my tools. Linux is very valuable and appropriate here.

At home, I don't have new problems. Chances are, I have problems that others have solved. I don't have time to roll my own solution. Nor the patience.

I want to take RAW photos and print them. I want to listen to music and maybe occasionally carry it with me. I want to watch videos on the Internet.

Yes. I know. Most of these things you can do in some form on Linux. However, F-spot is no Lightroom. Rhythmbox is actually almost as good as iTunes, but there's no Rhapsody equivalent. Flash videos sort of work (and they work on x64 if you try really hard), most other videos don't. †

Sure, free software is great. It lets you do so many things without paying a single cent. If you want to do something that's never been done before, it's a good place to start. But when you want to start doing things that it can't do already, and you don't want to spend many hours coding your own solution, that's where you hit a wall.

Why is this the case? As my friend always laments, "there's no good way to ship proprietary software for Linux."

Think of all the successful closed source desktop applications out there for Linux. Let's see, Bibble Pro, and VMware, Flash, and ... Real Player? Sorry, I've been using Linux for a while now, but I can't think of many more.

Bibble Pro works because it uses qt, and ships it's own qt library. If you're going to ship a C++ app on Linux, you better be prepared to ship every library you depend on. Trolltech's licensing sanity allows this.

VMware. well we spend lots of effort making our Linux product run on as many different distro's as possible. And trust me, its pretty insane. Just look into the vmware startup scripts if you're curious.

Flash? Well, Flash works because once upon a time a commercial company designed a plugin interface that was relatively stable, and has been good enough for all this time.

Real Player... didn't that use Motif?

Whether it's free software zealotry, laziness, or just the boringness of the work involved that leads to this situation, I know not. But in the end, without a better environment for proprietary software ††, closed-source thick-client apps will continue to be scarce on Linux desktops.

Without closed source apps, there's no proven business model. Without a proven business model, there won't be many businesses. †††

Without businesses, well there's no one that I can easily incentivise to solve my problems for me.

†Yes. I also know that a big part of why a lot of these don't work are DRM and proprietary codecs and the sort. Yes. I avoid these where possible. But these strategies aren't going to go away. Not until someone figures out how to really make money without them. And don't say Redhat. Look at how much they make compared to some software company that actually makes money. And don't say Google, because Google has just figured out a new way to deliver you their proprietary service and software that seems more open.

†† Ironically, the best support for closed source third party apps that OSS has produced is a reasonable HTML and Javascript environment.

††† Sure, some business innovate. Google is one of them. But there are very few Googles. And besides, if we do become totally web-based some day, who cares what OS you're running?

Update: I guess today is the day to chime in about Linux on the desktop. Anyhow, this artcile points out an important distinction. For many non-power users, Ubuntu has reached the level of usability and "good enough." These users will likely, as the article predicts, move more and more towards Linux. But its all the non-typical users that I'm really talking about. The gamers, the photographers, the artists, etc. There are too many commercial tools out there that are too hard to ship on Linux, that people will likely move to the Mac before they opt for completely open-source tools.

aKademy, the yearly KDE developer get together happened, and they fortunately put up all their videos of the talks. Fantastic.

Each video however is so fixated on the presenter, that it cuts out the slides!. And worse yet, not all the talks have their slides posted.

The two most important parts of a slide based talk, are the talk, and the slides. Sure it's nice to see what Shuttleworth looks like in action, but I don't need to watch him for 30 minutes straight, as I try to guess what his slides said.

I inadvertently ran a red light today. Yea, it happens. I've probably done it maybe 5 times total in 10 years of driving.

The problem is, the intersection that I happened to cross (Park Presidio and Geary) has a camera that captures you as you go by.

I'm probably not going to fight the ticket, but in my defense I'd like to point out the extremely poor design (IMHO) of the lights at that intersection.

If you've ever driven on Geary, you know that there are tons of lights. And they are all overhanging, in the middle of the road. At night particulary, you see a long row of lights stretch out in to the distance, and they are all fairly close together.

Except for Park Presidio and Geary. Here the lights for the east-west directions (on Geary) are the corner-lightpost style. So if you've been driving down Geary for a few blocks, and you're expecting lights in the center of the road and above you, you will likely not see these lights. Furthermore, because the lights on Geary are spaced closely together, it's easy to confuse a light slightly farther down the road as the next upcoming one.

That's exactly what I did. As I was going eastward on Geary crossing Park Presidio, I saw the 12th avenue lights and assumed those were the next ones I would have to worry about. And they were green.

Worse yet, I happen to cross exactly at the moment that the real lights had turned red, but the Park Presidio traffic had not yet started to move. I'm sure I wouldn't have driven straight into an intersection that had six lanes of intersecting traffic.

Furthermore, the camera that actually captures you is hanging from above! Why can't they just put the damn traffic light up there as well?

A super short review.

I finally went to the SF Apple store to play with one.

It's pretty good. Things seem to work as advertised. Typing didn't seem too bad, though I can understand how people would miss their physical keys.

I tried to post on this blog with one, but Safari crashed three times in a matter of ten minutes accessing different parts of this site.

Looks like they've got a bit more polishing to do. Seemed like something I could buy. But the Blackberry Curve looks a little more attractive for now.

I've been using the "classic" windows theme for a while now, but yesterday I decided to switch back. Big blue title bars and task bar.

Yep. It's ugly. But I find it more usable. There are two key features:

• The focused window is easer to differentiate. The main reason for this is that the window border get's highlighted in the XP theme, where as it doesn't change in the classic theme. In the classic theme, only the title bar changes.

  Of the XP themes, the blue one has the most contrast between foreground and background windows. Otherwise I would have gone with the silver.
  



• Bigger + Highlighting minimize/maximize/close buttons. Minor, but buttons highlighting as I roll over them seems to really keep me from missing. It's just that slight visual indication that confirms your intention.

Actually, this was mostly inspired by my realization that Metaciy is pretty good in this department. Especially if you pick a theme that lets you adjust the color of the foreground window.

Damn, they already even have a website.

This is Havoc Pennington's "vision" for Gnome.

I like the old saying about how those who do the work get to decide.

I'll reserve judgement on this whole online desktop thing, but what I don't want to see is a bunch of code written in a way that makes it hard for others to leverage GNOME and build new things.

While I don't mind people having vision, open source is about allowing different people to have different visions, sharing visions, and most importantly, working together on pieces that are common among these visions.

If anything, the GNOME project should be worrying about the boring stuff.. allowing 3rd party apps to integrate better, ABI compatibility, providing more reusable libraries as part of the core desktop, providing more wrappers to make programming for GNOME easier.

Most likely, the next great OSS innovation will come from a small group of people who get together and write some code to implement a new idea. The idea and/or the implementation will happen to be really good, and the product of this work will spread and be re-used.

The less code that such efforts have to produce, the faster the open source platform will evolve. The last thing we want is to have the limitations of a single vision causing these innovators to look elsewhere for a platform to build on. Maybe someday it will get to the point where combining available components will be easy enough that a small company could realize their single vision in the form of a distro.

So while Havoc and crew can go and make the "online desktop" possible, I don't buy into the single vision argument, since it seems to go against what actually makes open source software interesting.

There's a pretty depressing interview with famed kernel developer Con Kolivas and why he quit working on the kernel.

It yet again destroys the myth about how Linux is built on volunteerism and that it will naturally provide benefits to end-users solely because of its open development model. The interview bitterly points out that this cannot happen when all the big developers are being paid by people who care only about server performance. (I guess it's just a matter of who you define your end users to be)

I think it ties in a little bit with my previous post about how the Linux desktop needs to support commercial apps. The Linux desktop won't become dominant until someone figures out a way to make money from it. Maybe people can just make money by selling desktop hardware running Linux, but it seems to me that there's a lot more money out there if only people could easily sell end-user-oriented software running on top of Linux. If you get a big software vendor involved, they're bound to hire a few kernel developers to make sure the desktop experience is improved.

Open source is good (and the desktop experience is getting better), but money still makes the world go round. There aren't enough Con Kolivas'es in the world to get it done purely through good will and "fun".

Despite starting an Ubuntu tips site, I've been trying out using windows at work. The main reason? Fonts.

Yes, fonts can look pretty or ugly, but there is one major advantage that Windows has with regards to fonts. That is: small fonts look good.

At work, the more information I can keep on my screen at once, the better. I use a small programming font (proggy clean), which works fine on both Windows and Linux. What isn't available on Linux, however, is a small, readable, UI font. Having such a font can make a huge difference, as it reduces the space needed by menus and the like, and also allows lots of information display in applications like a mail reader.

So last night I tried to see, again, if I could win this battle. The goal was simple, find an 8pt UI font and font rendering settings that I would be happy with. I tried a long list of things:

• Various hinting settings
• Luxi fonts from the ubuntu 't1-xfree86-nonfree' package.
• Patched freetype/cairo/xft libraries that you can get from Ubuntu forums
• Redhat liberation fonts
• Lots of different fonts.conf settings. Turning off anti-aliasing, doing things different based on weight or font family.

It was pretty much all in vain. After hours of fiddling, I still couldn't come up with a configuration that I liked as much as the Windows default. That's just too sad.

But I did find some intersting links out of it, which I present here.

• A paper from the Anti-Grain Geometry project that has a very in-depth survey of different font rendering techniques


• The whitepaper for the "FontFocus" technique of making small fonts readable. Apparently this method was described a while ago, and its now unpatentable?


• Daring Fireball article about the changes in Mac OSX 10.3 that add a slight amount of vertical pixel alignment.