Submitted by Kaj de Vos made two big announcements at SylCon 2007. The first was a new web browser, based on a port of WebKit. The second is Syllable Server, which will bring together the Syllable GUI with the Linux kernel to create a server operating system that compliments Syllable on the desktop. Syllable Server is not based on any existing Linux distribution and will look and feel as much like Syllable as possible.
L A M E
the server dealy seems kinda neet but it makes me wonder if they are spreading themselves a bit thin. after all they have only a few people. i would personaly put as many resources into being a great desktop os. but thats just me
I wholly believe the focus of any new OS effort should be on userland/GUI with a stripped and much tailored Linux/BSD kernel below (Edit: NOT A DISTRO! Think more like what Apple has done with OS X)
A Linux/BSD kernel would offer a robust, secure, drop-in core and allow the majority of the effort to be focused on the creative aspects of OS design. I understand that it is a “learning experience” for those involved, but it’s unreasonable to hope to compare to what Linux/BSD have already achieved. I hope that after this server OS is developed, the Syllable team will see this as the direction to take the desktop and abandon the ATheOS core. If nothing else it will make the project much more realistic. IMO, anyway.
And why have I never heard of DirectFB!
Edited 2007-06-28 21:28
We will not be dropping the Syllable Desktop kernel for Linux. Syllable Server is a complement to Syllable Desktop, not a replacement. The Syllable kernel is great for desktop users, Linux is great for servers. The inverse is not true.
Huh? Why Linux for Syllable Server? Why not OpenSolaris?
I’m not trying to troll, but IMHO Linux is somewhere in the middle between server and desktop. (Open)Solaris — as its state is right now — is probably the best open source server OS available.
I can’t speak for them, but I’m betting they like the larger community and better hardware support.
Why do you think OpenSolaris is so much better for a server, just DTrace and ZFS or something more?
Becaue the Linux kernel has much more driver than OpenSolaris I guess?
Even for servers, this matter..
Things like hardware support, ease of integration and licenses (Both Linux & Syllable kernels and userland are mostly GPL & LGPL) are the main reasons. DirectFB is also a pretty good one.
What if the OS you have in mind wants to do things differently than Linux/BSD? Why would you be interested in them, then?
They clearly state in the story why they are not so motivated in abandoning Atheos core and choosing the Linux kernel instead:
“I demonstrated the Linux running on my laptop. It’s a humble machine eight years of age, with a 333 MHz Pentium III processor and 192 MB of memory. Our Linux only has a command-line environment so far. On that old machine, it starts in just under thirty seconds, whereas Syllable boots to a full graphical desktop in 17 seconds. On my development desktop machine, these numbers are 20 and 8 seconds, respectively. This should give you an idea why we developed Syllable in the first place, and why we are continuing to develop our own desktop kernel. On that laptop, XUbuntu based on Edgy Edge, which should be quite fast, boots to the XFCE desktop in two minutes and ten seconds…”
Edgy Eft is still using the old SysV boot-up scripts. Once Ubuntu fully switchs over to upstart scripts (even in Gutsy they’re still using SysV) there should be a nice improvement.
And I still think that this is wrong: the initialisation time given here is the total initialisation time not the initialisation time of the Linux kernel.
Using the Linux kernel doesn’t in any way force you to keep using the regular initialisation system of a typical Linux distribution.
Of course it’s slow, usually, it is based on shell scripts, sequentially started even!
I’d say that replacing the Linux OS initialisation system is much more easy than recreating a kernel from scratch and rewriting hundreds of drivers..
We plan to do this, eventually.
Yet we have a desktop kernel that works and drivers which support a good range of desktop hardware, so it is clearly possible. Besides, there are other benefits to having our own kernel, as well as historic reasons, all of which have been covered before.
Syllable was an OS focussed towards desktop users. Syllable on servers makes no sense. Most Unix/Linux servers are administraded without GUI anyway.
Syllable on Linux has IMHO some major problems:
1.) Current Linux users will most likely have an “Aw, yet another toolkit?” attitude towards Syllable.
2.) Common users (the target group for the Desktop release) will be confused. The Linux based (=Server) release will offer better hardware compatibility, but (maybe) harder configuration. If Syllable doesn’t support some kind of obligatory universal binary format (only apps that run on both kernels are allowed to call themselfes “Syllabe compatible”), Syllable user will be confronted with a bunch of apps that only run on either kernel.
3.) Fragmentation of delevopment ressources. Syllable doesn’t have that much developers anyway. The Syllable developers have to port every piece of code to both kernels and have to test their apps on both kernels.
I also have the impression that Haiku has overtaken Syllable in the “Best hobby OSS operating system” department.
The Syllable API already exists. Syllable Server doesn’t make this problem worse. If anything, it helps because we can have a wider audience for the API now.
It’s not as bad as you might think. We wont need to “port” the applications. We just need to maintain our lower APIs on both kernels.
Heh… Nice images from that SylCon
As to Syllable, it is quite amazing what that small team of developers has achieved in just five years. Congrats!
In terms of development, usability, available apps etc. Syllable has definitely been the alternative free desktop operating system for me (I consider Linux, BSD, Solaris and others like them mainstream, not alternative OS’s).
As to competitive alternative operating systems, Haiku may be promising too. However, as I never much used BeOS, and so never actually joined the fanclub I don’t always perfectly understand the meaning of trying to recreate and almost imitate an old OS (ok, the old BeOS GUI, including icons, looked very nice), however advanced it may have once been, years ago.
Syllable, on the other hand, seems actually more innovating to me when I compare it to Haiku (or maybe I’m wrong?).
Edit: As Syllable’s quite fast and its hardware requirements are very low, it could become the OS for older computers too (if there are working drivers for the hardware, of course).
From the article:
“Syllable desktop takes only a few seconds to start”
“it should be usable with 32 MB of memory, and quite comfortable with 64 MB”
Compare that to almost any Linux distribution with an equivalent GUI/desktop.
Edited 2007-06-28 21:57
As to competitive alternative operating systems, Haiku may be promising too. However, as I never much used BeOS, and so never actually joined the fanclub I don’t always perfectly understand the meaning of trying to recreate and almost imitate an old OS (ok, the old BeOS GUI, including icons, looked very nice), however advanced it may have once been, years ago.
b-because…. SHUTUP!
no, really, I loved BeOS and the way it forced things to be multithreaded and how it gave much higher priority to GUI related threads meant you had one hell of a responsive desktop! And by virtue of its design, beos spans multiple cpus really easily and actually USES all the cpus.
So having a new OS based off BeOS with lots of updates and enhancements is just a dream come true to some people. I really can’t wait for Haiku to hit R1 and definitely R2. I switched over to other OSes because of some limitations in BeOS that were annoying me.
I hear you. Firefox’s memory leak is super annoying. I went on #Haiku to ask if there was a fix or a way around it and they said “Don’t keep it open all the time.” ummm well, that’s not the ideal solution.
“So having a new OS based off BeOS with lots of updates and enhancements is just a dream come true to some people.”
Ok, as BeOS was a very good and fast desktop OS, especially suited for multimedia etc, I certainly agree that it makes a lot sense to study its basic ideas of multithreading and “how it gave much higher priority to GUI related threads” etc. and maybe even try to reinvent some of those things in Haiku. However, BeOS derivatives often at least give the impression that they “copy” a bit too much instead of innovating, and with “copying” I refer especially to GUI things and some other perhaps-not-so-relevant things that have not much to do with desktop responsiveness, multiple cpu support or speed. But like I said above, maybe I’m wrong, and I’m far from Haiku/BeOS expert?
Anyway, yeah, I guess that Syllable and Haiku do have different goals after all so comparing them doesn’t do justice to either one.
Edited 2007-06-29 00:17
I think you’re wrong. IMO they avoid copying, except from BeOS.
Go read some blog-posts over at haiku-os.org and I think you will see a bit of innovation. For instance
a) very compressed vector icon format. So small that you get them for free when reading the directory from disk. (Great for fast desktop filemanagement)
b) A new network stack. Not a port, hopefully with the ability to use rebuilt FreeBSD-drivers.
c) The interesting progress in getting a new schedular
d) 3d-driver for Nvidia cards, which was a one inspiration for http://nouveau.freedesktop.org/
(It was based on Utah-GLX though)
and much more…
There’s not much practical dichotomy between Haiku and Syllable. IIRC, AtheOS (the OS that Syllable was forked from) was heavily based on BeOS – very similar API, nearly-identical filesystems, they share a “kits & server” model, etc.
I’ve been following Syllable with interest (though I haven’t actually successfully installed it). But I’m concerned that adding yet another Linux distro to the hundreds already available may not be valuable enough to justify the danger of spreading the developers too thin to do a good job with the desktop.
It won’t be a distro. Distro means a group cobbles together existing projects into an ISO. This will be (from what I understand) Syllable APIs and the Syllable UI, but with the benefit of Linux’s featureful and secure kernel and driver availability. It really would be the best of both worlds.
Edited 2007-06-28 22:01
Isn’t this actually what Cosmoe was all about – the Atheos/Syllable GUI on top of a Linux kernel?
Anyway, the amount of work that already went into Syllable Server is amasing.
But I could see many people would use this system also as a desktop OS …
Great job guys/gals.
No, Cosmoe was (is?) an implementation of the BeOS API on top of GNU/Linux.
Stupid, half-off-topic question, but is there a plan for implementing suspend states / hibernation to the desktop kernel? The functionality is available in Linux (the server kernel), though it’s hit-and-miss. That’s a big deal for the laptop market.
Yes, work is progressing on adding the necessary support for hardware suspend and resume. The driver API already includes suspend & resume functions and some drivers implement them, and the ACPI bus manager is capable of calling those functions in the drivers. There is still a lot of work to be done, but the intention is there.
The latest builds also have support for ACPI CPU frequency scaling, and Tim ter Laak is working on an AMD PowerNow driver too.
That’s great news on the web browser. If Syllable has good enough applications for daily use, I might consider giving it a shot as a real operating system.
As far as the server goes, doesn’t sound like such a great idea. What’s the point of the server if it’s based on Linux. What’s holding the Syllable kernel back from being a good server OS?
What’s holding the Syllable kernel back from being a good server OS?
I imagine everything about it is optimized for latency rather than throughput. That results in a more responsive desktop, but lower performing on server type applications.
It would be nice to have an official statement of exactly what they hope to gain, though.
It should at least make a passable webserver since if I remember correctly, the AtheOS website ran on AtheOS for quite some time.
Linux is an excellent server OS. We’ve always said that Linux on the server and Syllable on the desktop would be a good combination.
Linux has the sorts of features and hardware support a server operating system needs. Adding these features to the Syllable kernel could be done, with enough time and developers, but then we’d just get another Linux kernel. So we may as well just use Linux.
We wont drop the Syllable Desktop kernel for Syllable Desktop because it’s been designed to do the job of a desktop OS, and it does it pretty well.
As a casual atheos and syllable user I feel that the priority should be on the application side rather than on diversifying the platform.
The main obstacle holding me back from more regular use of this OS is that is has almost no apps, resp. that the existing ones are often limited to very basic functionality and miss features that you’d expect nowadays.
Therefore, the news of the upcoming browser based on webkit is excellent. khtml may have been a very advanced project, but Apple has turned it into something a lot better! Even the KDE devs are planning to replace KHTML with Apple’s revamped version.
However, concerning the server plans I just have to join the choir of comments saying “oh no!”. The server marked is overfull with excellent OSs which do the job. OK, a server with a really intuitive interface which is really easier and more secure and featureful than others might have a chance, but do you really think, you have the resources to achieve this?
Please: Focus on improving the desktop. Focus on developing more and better apps. E.g. a mail client with IMAP support is a must. Just as a scheduling application, a word processor and a spreadsheet, which handle those infamouse de-facto standard formats from redmond.
ATM, syllable has a very nice and fast desktop, but there is hardly anything you can do with it, except for booting it, look around and switch it off again. This has to change.
Why is the idea of Syllable Server so hard to understand for people here?
Don’t think Unix, think Windows. When Microsoft produced Windows NT it did so because it knew that it would be advantageous for a Server/Workstation OS to look and act similar to the Desktop OS. Now Microsoft is working on a Home Server version because it knows that more people and small business (think one man) have multiple PCs and could benefit from a server.
The same holds true for Syllable. If you have multiple Syllable PCs running it would be nice to have a server, if you need to have a server it would be nice for it to look and act like Syllable on the desktop.
Making the Syllable kernel a good server kernel is possible but do you really want that if it means sacrificing features that make it a good desktop kernel? Besides, looking at how long it took for Linux and Windows NT to become decent server Operating Systems leads me to believe you need about 10 years for that.
There are also tradeoffs in using the Linux kernel instead of the Syllable kernel that make the resulting OS less desirable as a desktop.
* The result will be ABI incompatible.
* Startup and shutdown will be significantly slower.
* DirectFB is not the type of graphics subsystem suitable for a desktop OS.
Syllable Server is the best solution for somebody running Syllable on their desktop and needs a server. It is also a good solution for Windows users that want a Linux based server with good GUI.
However, at this point Windows was already a well-established desktop OS with lots of apps. Syllable has almost no apps, therefore it is of very limited use on the desktop. I feel that this would have to change first.
Edited 2007-06-29 13:46
Syllables user base is currently small, and Syllable Server doesn’t change that overnight. But they can be developed in parallel very easily, so there is no disadvantage in doing so.
Syllable really is one of the cooler OS projects out there and an upgraded web browser is a “Good Thing” I’m not really certain if Syllable for Linux will ever really take off, though… Even alternate GUIs which work on top of the X11 system seem to have a hard time gaining traction (GNUstep, anyone?), but X11-replacements seem to have an even worse track record…