February 13 Haiku was presented on Google Tech Talk at Google’s headquaters in California, and a special guest appeared on the event, the former Be Inc. CEO Jean Louis Gassée. He didn’t only join the Haiku team for their presentation, but also gave a few words of support and encouragement for their project. I, as many others interested in Haiku, found it great to too JLG’s presence, together with a few other ex-Be engineers (some of them now works for Google). The presentation was about an hour and is available at video.google.com.
The Haiku project is now about 6 years old, and is still not available in version 1.0. The good thing is, that the release is getting closer and closer. There were a few things in what the project leader Michael Phipps said, that I found not be completly true and I also saw one claiming the same over at slashdot. The first claim that I stopped on was: Compatible with Beos R5. Why? Well, this is both good and bad. When you start writing an OS more or less from scratch (the kernel was already existing), you need milestones and goals. Replicating BeOS R5 was on the paper a great idea, even I agree on that. But, and there is a big but, Haiku has untill now not had many active developers, and I think that most people involved in the beginning thought that Haiku v1.0 would already have been done by now. In fact, they still have at least a year to go! The problem is that the concept behind BeOS and Haiku will be terrible “outdated”, in terms of competing OSes having picked up the things that made BeOS unique. What will distinguish Haiku from the rest of the alterntives to Windows? We are also using computers in a different way then back in 2000.
Small footprint, fast boot. Yes, 16 MB uncompressed for an OS is quite impressive, but Linux even with X can be as small as 6MB uncompressed. In fact Linux is in many cases, prefered for embedded devices. Four commercial examples are Qtopia, Maemo, OpenMoko and ACCESS Linux Platform. There is even a few community driven embedded Linux distros. Second part of this is the fast boot time. Of course it is an advantage, or at least something nice. But in the corporate world, we rarely turn off computers. Updates and maintance of the computers are done during the night. More important is power management in this matter, when to force computers off to save energy and when to force them on to receive system maintance. Haiku will be lacking this functionality completly in version 1.0, and is hiding this behind fast boot time.
Unified, cohensive interface integrated, simplicity is the key, best defaults. Windows Managers, like KDE and Gnome, are not a bad thing. It means that the endusers have a choice and it can be described as one of the keyfactors for Linux success. These arguments are also often used by Gnome developers as “Less is more“. Something we all got reminded over this weekend! A small quote from Linus Torvalds is in place:
I’ll tell you why: because GNOME apologists don’t say “please send us patches”. No. They basically make it clear that they aren’t even *interested* in fixing things, because their dear old Mum isn’t interested in the feature.
It is often a question of going mainstream or being a choice for power-users. I guess there is no right or wrong on this, it’s just me loving simplicity in the way BeOS has provided me through the years. That is also the reason I love Ubuntu, it uses Gnome in a great way.
Less Debug – no need to test with FreeBSD, OpenBSD, NetBSD, Multiple Linux kernels, Windows. Again, Michael Phipps must have been talking about Windows Managers. I don’t know how many of the Gnome developers are working on a Windows version. I would guess no one! Some do work with a GTK package so that some applications can run on Windows. I asked a KDE developer yesterday how many KDE developers where working on other platforms then Linux. The answer was not many, and that KDE was optimized for Linux and that the KDE user experience was not optimal on other platforms. Now these days we see some screenshots of KDE applications running on MacOS X and Windows. That doesn’t mean that the whole Windows Manager has been ported. It means that some KDE developers have been using the official Qt package (made by Trolltech) to run a few KDE applications on other platforms, very much in the same way as with the GTK package. Now is this a problem? My answer is that this would only be a problem for Haiku. Why? The project only have 10-12 developers. KDE and Gnome have 100s or 1000s of developers and therefor can afford these kind of sideprojects. The sideprojects even bring something positive, they are tasters/advertisement of what the KDE and Gnome Windows Managers could bring the testers if they migrated to Linux.
Discourages forks and alternatives. Heh, if I understand this right, why is there a ZETA? Isn’t cause some people didn’t want to wait for an open source BeOS R5 in the first place? This sure was the case for me. What changed my view on this, is that I don’t want anyone to own my hobby OS. Hasn’t there been unsuccessful forks/alternatives of the Tracker and the Deskbar?
Human Interface Design. This is of course not something unique for Haiku. Again, this is not a problem achieving when you have a dusin developers, and that is why the Linux platform has freedesktop.org.
What I will give Michael Phipps right on, Haiku will not be bloated with libs. Linux has a huge problem here, but my view is that this is slowly changing. The KDE and Gnome develolpers are more and more working together through initiatives like freedesktop.org. An example Freedesktop’s Telepathy. It is a framework that has been adopted of both KDE and Gnome for all forms of real time conversations, including instant messaging, IRC, voice calls and video calls. Telepathy is making use of another thing that has become a standard in Linux, and that is the DBus messaging system to provide a simple interface for client applications. These initiatives are of course making Linux less bloated, but they have a long way to go!
Database like queries to find stuff. This was one of the things that made BeOS unique. But as mentioned earlier, as time has gone by other operating systems have gotten their alternatives. And I am the first to admit that I simply love the way Apple has implemented their alternative, Spotlight. Windows has Google desktop, Linux beagle, and there are a few other alternatives around. But, Haiku still has one of the best query solutions and this is the main reason for me having an interest in the project. The best potential of the OpenBFS filesystem is shown with the BeOS IM kit, a project I took part in.
A last word on the presentation. It took place to early. To attract developers to the platform, you should at least have a natively working GCC compiler. Another thing to have in mind for future presentations, make sure to only show things that works. Showing Firefox when you know it doesn’t work, is a bad move. Hopefully Google will let Haiku take part in the Summer Code, and of course I hope one of the projects then is the implementation of BeOS IM kit and work on OpenBFS.
Recent Comments