Linus Torvalds, creator of Linux, works at the Open Source Development Labs (OSDL), and is in a position to influence the future direction of Linux. So when OSDL's marketing director Nelson Pratt landed in London this week to talk to developers, we grabbed the chance to get the latest scuttlebutt.
Q: What are the big issues at the OSDL right now?
A: We see Linux going further into the enterprise but one of the big inhibitors is licensing. We know from talking open source customers that licensing on a large scale is too labour-intensive. The typical open source licensing granting process was set up with the view of protecting developer/hacker.
But today, as Linux gets deeper into the enterprise, it has to look back as a user of that software rather than a developer. We'd ask whether a licence was developed in order to more aggressively deploy Linux.
Q: How can the OSI make that happen?
A: We hear that the burden of reviewing licences is an inhibitor for Linux yet in two years it will be in 25 per cent of all servers. We want the OSI and other concerned bodies to see what they can do about that.
If you look at it from point of view of the customers, the odds are over 50 per cent that something will get done. Market forces dictate that, if an inhibitor is that licensing deployment is an issue, then it will be removed.
Q: What's OSDL's role in the licensing issue?
A: we want to be a voice of reason -- the place where vendors, developers and customers come together to talk after all. When asked, we can counsel but let me be clear that we have no intention of supplanting the OSI [Open Source Initiative] but we can be a strong voice in these organisations about that issue.
Q: As open source software usage grows and becomes more widespread, the forces on developers will change. Can the open source model of development scale?
A: Customers will out: in the long run, usage will point to the model that works. When it comes to open source, customers aren’t necessarily voting with their dollars but with usage, and most agree within the open source community that more customers is better and that non-technical inhibitors are a bad thing.
Q: But why would a developer write a device driver when it's much cooler to writing kernel code?
A: If customer says to a hardware developer that there's no driver so I won't buy your hardware, then the hardware vendor will write it.
I think the applications we see today will find their way onto most desktops. Even Microsoft codes for Apple and that's not a huge market, so it doesn't take a genius to see that plenty of developers will be coding for Linux soon if they're not already.
Q: What about desktop Linux -- the Holy Grail?
A: We try to get people to focus on the applications that Linux can do today. If a customer finds that Linux is not suitable, then they'll stay away from it, so a natural progression for Linux is for users to get comfortable with, say, a Web server, then for it to work its way deeper into the enterprise. That's how it will happen.
Q: How can you cut the price of support which most agree is higher for Linux than for Windows?
A: It will come down -- everything is expensive when new. But reports of high support costs are exaggerated. There's a lot of Linux skill dormant in the brains of those who have supported Unix for years.
It's not out of line with conventional desktop environments either -- we know of customers such as Unilever who don't have a support agreement with the distro vendor but the hardware vendor eg HP, Dell, CA etc.
Q: Will we see more consolidation in the Linux distro market?
A: We're focused on attracting developers, customers and vendors. We make no secret that we want to attract ISVs [independent software vendors] and customers, and both say fewer distros is better. ISVs say it costs more to support more Linux platforms, so it becomes expensive.
I believe market forces will come to bear. There will be more harmony and the market will make that happen. The differences will melt away. ISVs say to us that no-one wants one distro, rather they want their applications to demonstrate consistent behaviour across distros.
Q: So does that mean there will be a united Linux -- like the UnitedLinux project that failed when SCO sued IBM?
A: When SCO made the lawsuit, UnitedLinux stopped making declarations. There won't be a united Linux -- the goal is laudable but market forces will get us there.
Q: How is that likely to happen?
What's changed since the 70s, when there were lots of different versions of Unix? A: Then there wasn't a base of industry standard hardware, especially at the chip level. Today, every vendor has IA [Intel architecture] in their product mix, and software and hardware markets are more mature; in those days you had proprietary hardware vendors, which is no longer true. It's changed enough so that the Grand Unified Theory of Unix is Linux.
Q: Microsoft is a very successful company. What has it done that the open source community can learn from?
A: We could benefit from a unified view across distros, eg ways to consistently bring in ISVs so they can write once and run across different platform. There's not going to be a unified Netware or AIX -- software development is somewhat Balkanised, and the Linux development community would do well to address that, so they write once and have software run consistently across different hardware platforms.