Getting more applications on Linux

A highlight of the Linux Foundation Collaboration Summit (which I attended on June 13) was an afternoon panel entitled “How do we get more applications on Linux?”

Moderated by Dan Kohn, COO of the Linux Foundation, the panelist were Mike Milinkovich of the Eclipse Foundation, Darren Davis, from Novell, Kay Tate from IBM, Scott Nelson from RealNetworks, Ed Costello from Adobe, and Brian Aker from MySQL. (A word of caution: These quotes are based on my real-time notes at the conference, and I can’t guarantee that they’re word-for-word exact.)

I hadn’t expected that the panel would be dominated by a discussion of the Linux Standard Base, but it became clear about eight seconds into the discussion that the LSB is the end-all and be-all of getting third-party developers to target generic Linux.

Think about it: If you want to target Windows, you know which APIs to write to; if you want to target Mac OS X, you know which APIs to write to; if you want to target Solaris, you know which APIs to write to. Linux? You shouldn’t have to choose to write to Red Hat Linux, or SUSE Linux, or Debian Linux. You should be able to simply write to Linux.

That’s why the Linux Standard Base was envisioned as a set of consistent programmatic interfaces for applications developers. All Linux distros would implement the Linux Standard Base, and therefore, app developers could just target the LSB, secure in the knowledge that their source code should compile to run everywhere without modification. Customers, by looking for apps certified to run against the LSB, could be secure in the knowledge that apps would run on their own preferred distro.

Unfortunately, said Kohn, “There are a few hundred LSB certified applications, and a couple of dozen commercial ones. Why is Linux not getting thousands of applications?” Compare that against thousands upon thousands of certified Windows applications.

MySQL’s Aker said, “If the question is, how to you get [Windows applications] ported to Linux, well, you don’t! You never win there. You win in getting Linux used in new development. If the LSB can push new development onto Linux, you’ll win. If you find folks who are doing Windows applications today and get them to write in the first place for Linux tomorrow, that’s how you win.”

Novell’s Davis agreed. “The low-hanging fruit’s already done. I agree, the next step is new application development. We have to get ISVs involved in targeting Linux as a platform. What comes through Novell are customer requests, and that’s all customers care about. sure, We want to promote Linux, but customers don’t look at it that way. They’re looking at a business problem, and then finding an application to solve that problem. you have to have those on Linux.”

IBM’s Tate said, “We have great Linux certification by distro, and that’s the industry expectation. But being able to do certification at the application level, and teaching customers to look for that, is something that the Linux industry isn’t used to doing.” And indeed, the various commercial Linux distros push their own specific certification programs.

Eclipse’s Milinkovich expanded on this: “What you want is applications running on Linux. It’s a virtuous cycle. It’s not about getting products on top of Linux — it’s about getting the bank teller using Linux at the counter. Make Linux more ubiquitous, so you have an installed base of machines that ISVs will want to sell.”

He continued, “You’ve already eaten all the early adopters, now you have to get to the mainstream. Developers aren’t looking at the [software development] technology you’re including, like gcc and gdb — they think you’re nuts! You need a culture change. You have to give them application frameworks that they like. You can knock Microsoft all you want, but many developers like what Microsoft gives them to write apps. There’s nothing like that in Linux. Sure, I have an agenda, the Eclipse Rich Client Platform, which does that, but you have to recognize that if you want app developers, the status quo won’t get you there.”

A challenge, agreed the panel, is that the LSB, as a platform, doesn’t appeal to developers who aren’t hard-core Linux enthusiasts, but rather are considering it as merely as just another target platform.

Novell’s Davis said, “I have sent ISVs out to the LSB Web site, and they’ve come back screaming that they don’t know how to use it. You need tools to make using LSB easier. ISVs do the math and look at the market size, and can pick one or two distros — and that leaves the other distros out in the cold. We need LSB to work to stop it all from being dominated by Novell and Red Hat.”

MySQL’s Aker amplified that point: “I think it’s gotten worse. Different kernels on different distros means lots of different bugs that we have to support. That’s hard to describe to a customer, when there’s a problem that’s specific to a specific vendor build. This is getting worse.” He said, “There’s no money to be made in making tools. Getting vendors to implement new tools is really hard. Eclipse is the only forward option for tools.”

The moderator, the Linux Foundation’s Kohn, admitted, “The biggest competitor to the LSB is the makefile. if you have an open source app that you’re releasing, you create a makefile for each of the distros, BSD, Solaris, and so-on. The user compiles and it just works. The LSB project is trying to make the world safe for binaries, instead of source code, which isn’t the sexiest rallying cry. But Linux has to be that way to become a modern platform.”

That’s true. If your platform requires end users to compile applications, you’re not even in the game. As MySQL’s Aker said, “People use the distributions, and don’t want to compile their own applications.”

That’s a problem that this panel didn’t even come close to addressing.

Another problem that wasn’t addressed is that, for the most part, the majority of desktop application developers don’t want to target Linux specifically – they just want to sell or give away lots of software. The economics aren’t there – unless they have a specific business reason for targeting Linux directly, they’d be better off focusing on platforms with larger installed bases, and which therefore present a bigger business opportunity.

As the Eclipse Foundation’s Milinkovich said (and this was easily the best quote of the entire conference), “If you’re a startup, and the only platform you’re supporting is Linux, it sucks to be you.”