Foreword — This interview with Opera Software CEO and co-founder Jon S. von Tetzchner covers the Norwegian browser company's history, products, and roadmap. It sheds light on the growing importance of Linux in the embedded market, and the growing importance of the embedded market to Opera. Enjoy . . . !

---

Q1. Hi John. We had lunch together about five years ago in Palo Alto. I tried to push you toward embedded. How have things been going since then?

A1. Yes, I think I remember. There's a lot of things happening in embedded in general, and a lot of this is becoming web-enabled. That's a positive trend that we think is only going to continue. The amount of projects that we're seeing, and the diversity of projects we've been involved in is huge.

Q2. How many of the embedded projects you're seeing are based on Linux?

A2. Considering both mobile and embedded devices, it's between 50 and 100. We're up to almost 50 mobile phones, for example. More and more of those are based on Linux. We've made deliveries to Motorola and ZTe for their mobile phones. We've done the Nokia 770. We do the Sharp Zaurus PDAs. We're doing IP television, bar-code readers, airplane entertainment systems, shopping carts — all these are Linux based, by the way.

We're definitely seeing a trend toward Linux. More and more of the questions we're getting are Linux-related. We have two SDKs, and both are Linux-based. And, if you look at the number of Opera developers working on Linux, they're probably our biggest group.

Q3. What contributions does Opera make to the open source and technical computing communities, including contributions to open standards boards like the W3C?

A3. We're very active in the standards community, in particular. HÃ¥kon Lie is one of the main authors of the CSS standard. We're also fairly active in the WHAT (Web hypertext application technology) working group, which is us, Mozilla, and Apple working on next-generation Web standards.

We don't like proprietary stuff. We're working with others with the intention of posting it, which we think is a better approach [than creating in-house, proprietary technology].

When it comes to open source, we're not so active there. We use a few components, and we communicate with the folks doing them. Any bugs or issues we find with them, or improvements we make, are sent back.

Q4. How many Opera employees are there, and how many are engineers?

A4. We have about 270 people, of which 170 work on programming, quality assurance, or documentation. We are a technical company. We're market driven as well, but we're driven most by where we think technology is going.

We're seeing a trend toward more and more Web technology, and we're constantly evaluating emerging Web technologies and choosing which to support. That's something we've stuck to, and that we believe is the right way forward.

Q5. …such as which Web standards can best support device interfaces?

A5. Yes. We've been doing device interfaces for quite a few years, on set-top boxes, as one example. Some time ago, we developed a technique we call “Javascript plugins.” It provides, at the system level, extensions to the Javascript language that allow you to control the device. Javascipt plugins evolved into “Opera Platform,” which is now part of our AJAX [asynchronous Javascript] platform solution. It's something we made some time ago.

Frankly, a lot of devices such as set-top boxes have user interfaces that are not that great. Part of that comes from having been done in binaries, which makes it difficult to update. Well if you're using web standards, you have more people that can contribute, and also you have an engine, the web browser, which is a tested medium. You don't have to worry too much about updates.

I mean, if I were updating the binary of, say, a million set-top boxes, like they do over the air, I would be dead-scared. A lot of things can go wrong. But if you have a web-based system, as long as the web browser in the device is operational, you should be able to update the system by changing a website.

So the time to market, system upgradability, quality of service, everything is helped a lot by having a good web browser there. Combine that with Linux, and you have a flexible system that can also potentially support older applications written in binaries. But a lot of the UI stuff can easily be done through the browser.

Q6: Can you tell us a little about the Opera SDK? Is it based on Eclipse or another standard tools platform? Any plans to support Eclipse?

A6. Eclipse is certainly something that we're aware of. We are moving the Opera SDK forward, and providing more and more tools, but at this point, it's mainly a set of APIs.

Q7. And those APIs are based on C?

A7. Yes. In the SDK, we typically provide a browser engine, a reference UI that demonstrates the different APIs, and, more and more, tools.

Q8. The browser in my Nokia 770 doesn't look like Opera, suggesting that the SDK allows users to create fairly unique-looking UIs. Is that right?

A8. We are very flexible, and customer-oriented. We believe we can make a great UI, and if a customer wants us to do that, we'd be happy to. But in other cases, the customer wants to provide the UI on top of our engine. They can decide to have the UI written in C, or they can actually decide to have a UI-less product. For example, for a television, where you wouldn't expect to have any menus. As long as we have Javascript APIs controlling the device, everything can be done from inside the browser, and you don't really need anything else. So then it's just a matter of providing some Javascript/DOM applications for controlling the device.

There's a definite trend toward doing things this way, and the browser is getting more and more features to support it. Opera 5.0 had SVG (scalable vector graphics) 1.1 Tiny, and we have extended on that in the Merlin build, and we will continue to expand on that. There's a lot of things like that coming.

Q9. What emerging Web standards and technologies do you find exciting or interesting?

A9. There's a lot of good work happening in the WHAT working group. There's WebForms 2.0, for example. A lot of the forms pages out there are a total mess. Forms code is from a Netscape 2.0 timeline, and is very limited. Because of that, you have all these different solutions out there using Javascript, and it gets to be very messy. There's a lot of code to do the same thing 10,000 different ways. For instance, if you tried to go to a travel site, you'll see all of them have calendars, and they're all different. They won't really work very well on different size screens, or the like. It's a problem.

With WebForms 2.0, you have more types like this where you can specify for instance that you're going to be entering a date, and leave it to the client more how it's going to look. WebForms 2.0 is an extension of what we've got, because we don't really think the Web as we know it is going to go away.

Then we have Web applications, including things like Canvas. Canvas makes a part of the screen a bitmap that you can draw into, inside the browser window. By using Canvas, you can make just about anything. You can use basic drawing functions and line functions, and you can think where that leads. You can do games in it, a lot of cool stuff.

We have SVG for 3D graphics, and then [with canvas] you also have a way to do the simple stuff. You combine that with DOM and you get quite interesting results.

Q10. DOM, as in “document object model?”

A10. DOM allows you to take any item on the screen, and do anything with it, in your document, whether it's visible or not. DOM is a combination of CSS and Javascript, in a way, meaning you can move things around on the screen. When browsers get DOM, it changes the browser quite significantly. It has taken a lot of time for each individual browser maker to get it right.

If you think about Netscape 4, and Netscape 7, that's before and after DOM. It means you go beyond static pages — from pages with only a few dynamic elements, such as animated gifs, or some scripting to handle forms. With DOM, anything on the page is something you can program. It's Turing-complete, and you can do anything.

Q11. Are there other browsers in the embedded space that have a DOM?

A11. There are players out there that claim to have one, but it's sort of like Netscape 4's claim to have CSS support. They had 38 percent CSS support.

It's sad but true, but most browser makers don't implement Web standards fully. It was a trend started by Netscape, but then followed by Microsoft. We try to do it, Mozilla tries to do it, but a lot of the other players don't.

Q12. CSS 2 is still pretty far away from being fully implemented by anyone, though, isn't it? Like the ability to download fonts into the document, and the voice reader stuff? Do you think that'll ever get implemented?

A12. I think that's CSS 3, but I may be wrong. You also have to remember that CSS 2 was replaced by CSS 2.1, which interestingly enough, is a subset of CSS 2.

We are more or less compliant with CSS 2.1. Anything that doesn't work there is a bug. But when it comes to CSS 3, we're already doing a number of things. I guess font-downloading hasn't taken off yet, but it's something that's of interest to us.

Q13. What about voice technology for devices? Does Opera support that on Linux, or only on Windows embedded operating systems?

A13. Currently, it's only being distributed in the Windows version [of the Opera browser], and actually you have to have XP or 2000 for it to work. I've seen it demonstrated on Linux devices, though, so this is something we can deliver, on the device side, if someone's interested in using voice. It's not there on the Linux desktop yet, but that's really dependent on IBM providing us with that library on the Linux desktop.

Q14. What about Opera Mini. Is that product available for Linux devices?

A14. Any device, basically. With Opera Mini, we run the browser on a server. We use a small screen rendering technology that we download onto mobile phones. We have two versions, one for powerful phones, and one for very low-powered phones. We spent a lot of time to make the Opera browser run lightly, so we can handle quite a few users with a small number of servers. It's still the biggest server park we ever made, but we are able to handle a lot of users. There's 700 million mobiles out there with Java capabilities that can download the small Java applet, and get the Internet.

Q15. Do you run a copy of the Opera browser on the server, for each Opera Mini user that's currently logged in?

A15. The details of that, I don't really want to go into. But we do have a lot of instances of Opera running on the server, yes.

Q16. Is service provider support required to use the Opera Mini browser, and how is the software and service priced?

A16. No, it just needs to be on the phone. The service provider may potentially get in the way, but I don't think there's too much of that. In some cases, users may have to pay their operator to connect to data services. Overall, we're trying to make it as simple as possible, and in a lot cases, it just works, out of the box.

We are providing the service free to end users, and working with operators who may like to have customized versions of this, which they then pay for. But the end-user is not paying us anything. They may be paying someone else something, and obviously they're paying for the traffic one way or another, but we don't get anything on that unless we have a deal with the operator.

Q17. Before we move into more general questions about the embedded Linux market, are there any Opera products or services you'd like to talk about?

A17. We've covered quite a lot. We are providing the browser on just about anything that flies — or doesn't fly. We're able to do that because we're able to get it working on a new operating system in a very short time. I think the record for a prototype is nine hours. Typically it's more like a couple of weeks. And sometimes, I can tell you, those are operating systems that are much worse to work with than Linux. Linux is a dream compared to a number of OSes we're working on. Obviously, they have a benefit in that they're more deeply embedded, but at the same time they're not typically used to run a complex application like Opera, so things like memory handling may not be in place, or might be fairly limited, or bad.

Q18. How would you characterize the current state of embedded Linux? Is it still in a rapid improvement cycle, or getting to the point where it's pretty good? And also, where do you think Linux is in the market, in terms of its acceptance?

A18. We're seeing a lot of interest in Linux, especially in the mobile phone space. You've seen that Motorola is spending resources there. Obviously you have the Nokia 770. How far Nokia will go with Linux remains to be seen, in a way. I mean, we have to be a little bit careful. Sometimes we know a little bit more than others, but we're definitely never allowed to talk about it.

We're seeing a tremendous interest in Linux from companies in China, Japan, and Europe. In the mobile phone space, things tend to take a long time. If you try to analyze the mobile market, there's always interest in Linux, and Windows, and Symbian. The smartphone segment of the market is still small, even though it's 700 million units. So, if you're only talking about a five percent marketshare, it's still very significant in terms of volume.

But we're definitely seeing an uptick of Linux interest in the mobile phone market. Symbian is the leader in smartphones, but Microsoft and Linux are having a nice little battle over second place.

When it comes to other devices, the situation is different. Our impression is that a lot of the projects are going toward Linux. There are other OSes around, obviously, but there just seems to be a trend towards Linux. There is also some Windows CE around, but the majority of the trend, from what we're seeing, is towards Linux.

So, my assessment of Linux in general is that it's a tremendous success on the server side. Linux in embedded is moving quickly toward success. It has already been used in a number of very successful devices, and seems to have a healthy pipeline. And then on the desktop, I think Linux still needs to live up to its potential.

Q19. What percentage of Opera's desktop browser downloads are for the Linux version?

A19. Probably something like ten percent.

Q19b. Oh, that much?

A19b. I think Linux users, more than anyone else, have a tendency to want to try out alternatives.

Q20. Do you have any comments on any legal threats to Linux such as software patents, DRM, copyright infringements?

A20. I think there are threats. At the same time, you also have some big backers, in companies like IBM, and others that are building solutions based on Linux. And some of those backers happen to be the biggest patent holders. That's kind of interesting in a way, that some of the biggest supporters of Linux also have some of the biggest patent portfolios! In a way, it doesn't match, but… that's the situation.

If someone wants to go after Linux, as a platform, they will then have to fight some of these larger companies. Obviously, that may work for someone, but it's going to be a tough battle, and if they succeed, there just might be a patent that some of the Linux supporters will find in a drawer [to fight back with].

Patents are a problem in general, and we've been quite active in patent discussions here in Europe. Whether to allow software patents is one question, and what kind of software patents to allow is another. There was one round that got won by people opposed to software patents, and then another round where we won again. We'll see what happens next. It's probably not done. There's some big forces behind getting tougher patent legislation in Europe. But so, it's been going the right direction.

Q20b. As a proprietary software company, wouldn't you stand to gain an advantage from software patents?

A20b. The question is what kind of company you are. We are for open standards. We would like to compete on the quality of the products we deliver. If there are to be patents, they should be fairly difficult to get, and actually have a significant amount of innovation. I guess that's our stand.

Q21. Can you share one or two of your companies most exciting successes?

A21. Obviously, the phones we've delivered, some of which actually bear the Opera logo. That makes us proud. Like the first phone in Japan that was Opera branded — it sold out in 24 hours, with what was supposed to be three months of stock. That was a good feeling.

We pride ourselves on doing well in standards tests. Something that made us proud in the old days was when we had our first CSS implementation, and Netscape was at 38 percent, and Microsoft was at 70 percent. We had 90-something percent compliance, in our first version. We spent three months doing it, and Microsoft spent three years. That's something that obviously made us proud, and that's actually the reason that HÃ¥kon Lie joined us.

He and I had been working together at Telenor Research, then he didn't really believe [that the Web was very good], so he went to the W3C, and wrote the CSS standard, while we made the browser. He worked with Netscape and Microsoft for three years to get them to implement it, and they had kind of halfway solutions. Then he spent three months with us and we had a better solution. That made him convinced that we actually could compete on an international level, which he didn't believe when we started off.

Another product I'm proud of is IBM ThinkPads. If you have a technical problem and Windows won't start, there's a rescue system. That rescue system is running Linux and Opera. I'm kind of proud of that; if Microsoft fails, you have a rescue system with Linux and Opera.

Q21b. Is pre-boot rescue a growing market focus for Opera?

A21b. It's basically something that IBM wanted to do… it's important that you provide good systems if there are problems, and Opera is a good choice there because we have an embedded browser that gives you a lot of flexibility, and works “light.”

Sometimes people tell me how powerful computers and networks are today. But they were saying the same things 10 years ago. And if you think about the computers and networks we were using 10 years ago, I don't really feel they were that fast. Do you? So our feeling is that continuing to focus on running on older hardware and squeezing the last ounce of power out of it is a good thing to continue doing.

Q21c. Opera's desktop browser is a lot lighter than other desktop browsers, but I imagine the embedded version must be even lighter?

A21c. It's the same browser, but we lost all the chrome. We're not including the mail client and chat client and Usenet client and RSS reader… or the installer. If you throw all these things away, you have an even smaller package. We have a build that can run from a floppy, and some of the builds we're providing for mobile phones have that kind of a footprint. If you give us compression as well, we [can definitely deliver a floppy-sized image].

Q22. How about a failure?

A22. Every time we're late with the desktop product, or any other product, that's a failure that I find very painful. We try not to be late, but we've had more delays on the desktop than I would have liked to see. We're about to hire a lot more people to make sure that doesn't happen as much in the future.

Also, we've made products for companies like Ericsson and Nokia that never got shipped. In a way, we didn't fail, but the products got cancelled, so that wasn't really too much fun. We don't like our people to waste time. In general, 95 percent of what we do goes into products, though, which I believe is a fairly high percentage.

Q23. What opportunities do you see in the embedded device market?

A23. Really all kinds of stuff. For example, if you work at IBM, you can see Opera running on a refrigerator…

Q23b. Really? To look up recipes?

A23b. It's kind of cool. It will actually recognize what's in there, and that kind of thing, with barcode reading and other technologies. I have to be careful not to say too much, but a lot of it is based on Linux, and they're using Opera.

Q24. What technical challenge does Linux face, and what technical possibilities in Linux do you find exciting?

A24. Linux is a nice embedded OS to work with. You have a lot of the functionality you're used to, and a number of tools available to you. On the negative side, you do sometimes notice that Linux isn't made for embedded, in a way. As one example, by default Linux will never return [an error] when allocating memory. It will try its hardest to make it work. Which means that it can become slower and slower and slower and slower, or eventually crash as you try to access memory that it didn't manage to actually allocate.

Q24b. The dreaded OOM (out-of-memory) error. There's some discussion about that taking place in some of the special interest groups for embedded, I think LiPS or some of the others.

A24b. I would presume this is something that people are trying to solve. We spent a lot of time making Opera memory-safe. It took us a long time, and was quite painful. That's one of the benefits you have with a very structured, organized team — we managed to solve it. But it was delaying just about every other project, because it affects just about every line of code. You need to check that you're doing things correctly. One thing is to just not crash. That's the simple thing. But you actually want the program to preferably not fail.

Let's say we are running Opera on a device that has something like 4MB of RAM for us to work with. The device may well have 64MB of RAM, but maybe we only have four. That means if you download a page which is half a megabyte to a megabyte, you're down to an out-of-memory, all the time. Then the question is, how do you handle that? Worst case, you crash. Best case, the user doesn't know you ran out of memory about a thousand times while downloading that page.

Q24c. We've been wondering, using the 770, how it does it. It's very smooth. We haven't seen any problems. We can't quite figure out what magic Opera is performing…

A24c. We spent a lot of time on this. I mean a lot of time. But, we are already running on devices that don't really have memory management at all, and we need to handle a lot of this stuff ourselves. At the same time, we also need to work with the rest of the system, so we can't really do everything on our own, either. So it's quite complex, but we've spent now seven years doing it, in the embedded space, and a few years more in the desktop space.

Q25. In a quantitative sense, how significant are embedded and mobile to Opera from a business perspective at this point?

A25. We've been [deriving] two thirds from embedded and mobile, and one third from desktop.

Q26. Of revenue?

A26b. Yes.

Q26c. What about profit?

A26c. We haven't been providing profit data from the different sections. The way we see it, they're intertwined. To be able to make a good embedded browser, it needs to be a good desktop browser.

Q27. We've been hearing rumors about companies interested in acquiring Opera. Is that something you don't prefer to see happen?

A27. Definitely. We worked a long time to build the company. We've had offers throughout the years. There's been a lot of rumors about us being bought by anything ranging from Google, to Microsoft, to — I think some of the guys at work were having a competition to guess 'who's next?' and they were wondering if the rumor's going to be about Oprah Winfrey buying us next time.

What we are interested in is getting as many users of our products as possible, and providing the best possible product. At the same time, we are also a public company, so we have to provide results to our owners. Which, by the way, includes a lot of employees, including the founders, which also includes me.

The two founders, Geir Ivarsøy and myself, we did have a lot of offers. We could have walked away with a significant amount of money, at the early stages, but we've been wanting to build the company. So, we have no plans whatsoever in this direction. We don't control it, totally. But for instance, selling out to Microsoft, that's… [laughs]

Q28. You've done really very well, and we congratulate you. And we look forward to seeing you swim across the Atlantic sometime.

A28. You never know, I might make another try of it.

---

About the interviewee — Tetzchner worked for Telenor Research from 1991 to 1995, when he and his colleague Geir Ivarsøy founded Opera Software ASA. He holds a master's degree in computer science from the University of Oslo. He also holds 17,482,110 shares and 200,000 options in the company.

---

 
This article was originally published on LinuxDevices.com and has been donated to the open source community by QuinStreet Inc. Please visit LinuxToday.com for up-to-date news and articles about Linux and open source.