"A Breakthrough for Linux..."

Linus Torvalds Talks about Current Linux Issues

Jürgen Schmidt
c't

Translation by Jürgen Schmidt

Linus Torvalds is still the main person in the Linux community. The "father" of Linux still coordinates the further development and is regarded as last resort when it comes to controversal questions. c't talked with him about current issues regarding the freeware operating system

c't: A couple of days ago, two major Database vendors - Oracle and Informix - announced that they are going to offer their products for Linux. Do you think this is a breakthrough for Linux in the commercial world?

Torvalds: I think it's a breakthrough, not so much because I think it's important for Linux to have Oracle, but because it shows how big an impact Linux has had, and that big commercial vendors are beginning to port simply because that's where a noticeable portion of their market is moving.
So I'll enjoy having Oracle for Linux, but I think the more important fact is not Oracle moving to Linux but so many others having moved to Linux that Oracle felt it was in their best interests to do a port.

c't: How is this going to affect the future development of the role of Linux in the computer world?

Torvalds: Well, for some people it will mean that Linux will become an alternative where it didn't used to be one, simply because those people required Oracle for their work.
The bigger impact I think is just the psychological and market impact of Oracle having ported to Linux - like the Netscape announcement it gives more credibility to Linux in the minds of lots of people - even if those people aren't necessarily actually even Oracle users. So there are secondary advantages, not just the availability of Oracle, but simply the added mindshare it gives Linux.

c't: You always propagated Linux as an operating system for the desktop. Now it seems as if Linux is getting most of it's public attention and credits as a server system. Researchers like IDC position it as an competitor for Windows NT not for Windows 95/98. Aand Linux made its major inroads into the market in commercial environments as a server. What do you think of this developments?

Torvalds: Note that the reason I propagate Linux as a desktop operating system is because I think that's the more difficult market. I was personally never very worried about Linux as a server platform - servers are "easy" compared to desktops, because in the end servers are fairly anonymous - you only see the network behaviour of a server, while with a desktop system it's much more of a "complete immersion" environment, not just the network part.
So I think that Linux is making more waves in the server area simply because it's the easier market to enter. I still personally consider desktops and personal computing to be the "final frontier" in some sense.
Don't get me wrong - I think servers are important. I'm just not as deluded about their importance as all the commercial UNIX vendors..

c't: What do you think of KDE?

Torvalds: I don't use either KDE or Gnome, but from what little I've seen of KDE I like it. However, all the flames in the net about KDE vs Gnome has turned me off both to some degree, and I hope that they'll start working together more.

c't: Are you using it ?

Torvalds: I'm still in the dark ages. I'm using a rather basic fvwm setup that looks much like the default installation of Linux used to look a few years ago. I'll eventually upgrade to a newer look, but as I'm working mostly on things that have absolutely no bearing on the desktop I haven't felt much of a need so far.

c't: There has been a lot of discussion about the licensing issues that arise from the fact that the basic GUI-Library Qt is a commercial product and not available under the GPL. Do you think this is a problem?

Torvalds: Yes. But I think most of the problem stems from people being so divisive over it, and flaming each other on the net.
I used to think that Qt was fine, and I still think it's a technically superior product with a good company behind it. However, these days I wonder if people can accept KDE due to political reasons, and that makes me sad.

c't: Redhat - one of the major Linux distributors which is quite dominant in the US - refuses to deliver KDE with their distribution. Instead, they propagate and actively develop GNOME. Do you think this is the "Right thing to do(tm)" ?

Torvalds: It's their choice. They do have some point, in that due to Qt licenses they cannot do everything they'd possibly like to do with it - if QT has a security bug the license would still prohibit RedHat from fixing it even though they have the source etc.

c't: A desktop environment with tight integration of applications is a huge project. Do you think the Linux community has the ressources to develop two of them in parallel?

Torvalds: It may be that the community wouldn't have the resources unless there were two of them. To some degree the animosity between the KDE and Gnome group seems to have spurred more development. I find the animosity distasteful, but it may be that whoever ends up being dominant will be technically much better for having had some competition.
We all know where a lack of competition can lead us...

c't: Are we going to have a "Redhat Linux" that looks and behaves totally different for the average desktop user than for example a "SuSE Linux"? Or, put the other way round: Is Linux going to repeat the history of Unix, which threw away its chances for the desktop mainly because when the Unix companies finally agreed on CDE, everybody was already using Windows?

Torvalds: I doubt it. Unlike UNIX, the tools still come with sources. Whether KDE or Gnome becomes dominant is fairly unimportant in the long run, simply because even if it's dominant people will still be able to look at what it does and how it does it and duplicate the functionality in the other. The Gnome people may not be able and willing to use Qt, but they can essentially easily duplicate everything else.

c't: Intel is already making a lot of hype about their new processor generation Merced. Commercial operating system vendors already have emulators that simulate its behaviour, and they are actively porting and testing their operating systems with them. What is the situation for Linux in this regard?

Torvalds: I personally don't want to sign NDAs, which means that I won't be getting Merced information before the release. However, there are others that are interested in Linux on Merced that are more flexible in this regard, and I'm sure we'll have a Merced port very soon after the release of the chip.
The biggest problem by far is the development tools - gcc and binutils. The kernel proper I'm not in the least worried about: other OS vendors may need simulators and several years, but Linux has both the PC side and the 64-bit parts handled already, so a native Merced port has no intrisically hard issues to deal with for us.

c't: Is Intel talking to you and other kernel developers about Merced and Linux on Merced ? When will you get the necessary informations as well as soft- and hardware to start porting Linux on Merced?

Torvalds: I've been talking to Intel, but at this point mainly about high-end server hardware like the four-way Xeon etc.
As to Merced, I can only say that I know of people working on it.

c't: Last but not least: 2.0 has been out for more then 2 years now and the developer kernel in 2.1 are already beyond 100. When will we see the next major stable version and what is it going to be - 2.2 or 3.0 ?

Torvalds: The next version will be 2.2, and code-freeze for it has essentially descended on us already. So I hope the final 2.2 will be out in a month or two.

c't: Will it bring major enhancements for the average user or will it deliver just support for new hardware?

Torvalds: Most people won't notice any major new features. It's a lot better on various non-intel architectures, and it's a lot better on high-end hardware - SMP and large-memory machines. It has some new features that will mean that it probably makes sense for people to upgrade from 2.0 even if they have "average" machines - better threading support, and Wine may need to run on 2.2 in order to do good Win95 threads emulation, for example - but I think most users care more about other upgrades - like upgrading their window manager etc.

c't: What are - in your opinion - the most exiting developments concerning Linux?

Torvalds: Most of the excitement is going on in user land - I personally get excited about Linux/SMP and scalability issued of the kernel, but I think that user applications etc will and up having abigger impact on where Linux will be in a few years. Only kernel oldtimers like me really need to care deeply about the kernel itself... (ju)

Last changed by Jürgen Kuri, 07.08.98

Copyright Verlag Heinz Heise