Linus Torvalds is rapidly becoming an influential figure in Silicon Valley. The Linux operating system he created continues to gain attention, and Torvalds now working at the enigmatic Transmeta Corp. finds himself guest-speaking at conferences or fielding questions from the mainstream media.
The key to Linux's popularity lies in its open-source model, which makes the source code available to anybody for modifying. That's meant that the OS has evolved as actual users tweak and twist the code around; it's also made Linux a favorite alternative for engineers fed up with Microsoft.
Applications are joining the Linux camp as well, with companies such as Oracle Corp. promising Linux versions of their wares. Even Intel Corp. got in the act, making an equity investment in Linux provider Red Hat Software Inc. Yet Linux remains a dubious and cultish presence in the eyes of many, and while EEs claim to be crying out for Linux-based design tools, most EDA companies don't see much profit to be had in that business.
EE Times recently asked Torvalds for his thoughts on the future of open software in corporate circles.
EE Times: What's going on with the road map for Linux that might be helpful to the high-end technical users?
Torvalds: We actually don't have much of a road map. Linux tends to be developed by reacting to what developers are interested in and what users ask for. [That method is] much closer to the user community.
Right now, one thing for high-end users is probably the next stable kernel, which should be out in a month or two. It's going to be very much SMP [symmetric multiprocessing]-oriented and very stable. We've had SMP support for a while, but it's been kind of an ugly stepsister. Now it's very much integrated into the standards, and that obviously helps for doing large engineering work, where SMP is fairly common.
EE Times: One thing that keeps coming up as I hear you talk at different conferences is the prospect for getting applications up for Linux, especially engineering and CAD applications. What are the biggest barriers to that?
Torvalds: Well, I actually used to think that the most important barrier was the perception issue, and I think that for various reasons the perception issue is it hasn't gone away, but I think it's a much smaller barrier these days. I mean, Oracle did a very big thing in turning people's perceptions around [Oracle Corp. in October released Oracle8 and Oracle Application Server in Linux versions], and once you have Oracle ported on Linux, suddenly people think: Wow, what's missing here? And then there's the support of companies like Intel that aren't exactly known for jumping on the bandwagon they're fairly old-fashioned, usually.
I think that perception-wise, we're in pretty good shape, and I think the major obstacle now is it's kind of hard to find cold, hard numbers of how many people are using Linux, because a lot of them are getting it over the Internet. That makes it sometimes a harder sell at some companies. And it's going to take time. Even when you port from just another Unix and porting to Linux tends to be one of the easier ports, according to all the reports that I ever saw there's still qualification, verification and support, which are big issues for a lot of companies. It's not something you step lightly into, even when it's fairly simple to do the actual ports.
So it's going to take time. But when it comes to the higher-end user market, like CAD tools and so on, where Linux is already a major factor, I think it's just a matter of time before you will have one or two of the CAD vendors starting to get Linux versions and the others will probably be forced to follow suit.
EE Times: Do you think Intel's support of Red Hat has helped out?
Torvalds: Yeah. I don't think it makes much of a difference in the technology sense. It's more a psychological edge.
EE Times: A lot of our readers say they run Linux on their computers at home. Has it made inroads into getting on their computers at work?
Torvalds: I actually think that a lot of engineers were introduced to it at work. I go to these conferences, and you have the show-of-hands kind of thing, and usually when you ask, "How many of you use Linux only at home," you get maybe 10 hands. And when you ask, "How many use Linux at home and at work," it's like, the rest. So I suspect that a lot of engineers use Linux both at home and at work.
EE Times: How do independent software vendors, the likes of Cadence and Oracle, develop for Linux given its open-source-code status? Do they just grab a copy and start playing with it, or do they want a contact at some place like Red Hat?
Torvalds: It really depends on the company. A lot of companies just decided to make a Linux port and sell it to people. What I saw with Oracle, for example, was that Oracle really did want to talk to Red Hat first. Not because they really needed to or had to, but simply because that's how they've always operated. There have been fairly close ties with the other Unix vendors, and they just wanted to feel comfortable about Linux and treat Linux as any other commercial Unix.
EE Times: What is the future of open-source software for engineering applications? Will we see more of them?
Torvalds: I think there's going to be more, but I don't see the kind of exclusive open-source mentality that some people have. I think that where you'll see it is, especially when it comes to infrastructure applications [i.e., the operating system], you'll see it in a lot of other things that people don't make money on, per se, but that they really depend on to get their applications working. Like compilers, networking substructures, whatever. I think a lot of those will be open source just because it makes a ton of economic sense to not have one company do them.