Emanuele from the Mantova Unix User Group in Italy had a chance to speak to Travis Geiselbrecht, the NewOS creator. Travis has worked at Be’s kernel team and he is now employed by Danger Research, while in his free time he is helping the Open BeOS developers to integrate his NewOS kernel into the OpenBeOS one. Travis is talking about his the future of NewOS, Posix and he is giving his opinion about BeOS and its future. Descriptive quote: “It’s pretty obvious to me that Palm is buying the engineering team of Be and I see absolutely no point in Palm releasing [BeOS] R6.”
Travis Geiselbrecht on NewOS and the Future of BeOS
Submitted by Emanuele 2001-11-25 Haiku 23 Comments
Its mildly interesting that the creators of the BeOS kernel don’t use it anymore. All the same, I think its their loss. With OpenBeOS and BlueOS both making headway, it won’t matter much if Plam releases R6 or not. I hope they do though, that would be awesome. I’ve used BeOS exclusively at home since around R4, and I don’t see any reason to quit yet. Lack of games is a bitch, but thats what a PS2 is for. Its still fast(est), friendly and stable – – that hasn’t changed one bit. Hopefully by time semi-modern hardware isn’t supported under BeOS, one of the OSS-BeClones will be usable or someone will have actually produced a good Linux distribution (good as in competitive to BeOS on speed AND ease of use, which is yet to be matched by ANY OS).
I see absolutely no point in developing an OS that runs on a dreamcast. Like NewOS.
“I see absolutely no point in developing an OS that runs on a dreamcast. Like NewOS”
So I’m assuming you wouldn’t want to develop for Linux or FreeBSD? They both have Dreamcast ports in the works.
It’s OK Fpouler, although we talked about making you “develop an OS that runs on a dreamcast” we eventually decided to let someone else do it. So now it doesn’t matter whether or not you see a point in doing it
The most I expect from NewOS is that Travis will find out for himself WHY this or that odd compromise has been made in the design of Solaris, BSD, Linux, NT and so on, in a way that one can’t without facing the same obstacles. He might even discover something new and interesting, or demonstrate something which was believed possible but never previously tried. More likely though NewOS will never actually continue beyond the “tinkering” stage, and Travis will be happy to let it go.
IMHO we don’t need another OS kernel to solve any of the problems pointed out by OSNews contributors and critics. The most interesting problems are in userspace, and will be solved there. For an experienced programmer the design and implementation of a kernel remains an interesting exercise though, and I wish Travis luck in persuing it.
I’m curious, how old was the interviewer? 5?
Do another interview better than the mine: i’ll be happy to learn how a real interview is.. Or you’ve still to understand what’s he difference between simple & stupid questions? Btw, 20!
A few quotes for background….
Hi Travis! First of all, thank you very much!
Hey, no problem. I’m just doing what I like to do!
I wasn’t very happy as a kernel engineer at Be.
eheheh… Which development platform you use? I suppose… BeOS.
No, not BeOS; Win2k instead (though I sometimes compile on a FreeBSD, Linux or Solaris box). I haven’t used BeOS in months!
Doh… you created it but you do not use it!?
Not anymore: I’ve moved on. Most people I know that worked at Be and quit never used BeOS again
Ah…. Well, strange isn’t it?
Okay, okay. I do like it, but it’s not useful for me anymore. I’ve started viewing the computer more as a tool than a religion, so BeOS isn’t really all that useful to me anymore. I’ve gotten really tired of figuring out which hardware works and which doesn’t! Now I can just use whatever hardware I buy, which is kind of nice. Windows has it’s problems, but all OSes do. People just tend to get “religious” about their platform and forget about the problems it has.
Ok… Let’s stay with BeOS. First of all, do you think that a R6 will come out soon or that Be Inc. will stop developing this OS?
No and yes! It’s pretty obvious that Palm is buying the engineering team of Be and I see absolutely no point in Palm releasing R6.
1) I’m pretty tired of BeOS. I’ve moved on and dont use it much anymore.
I don’t know if NewOS will ever become useful by itself, but it’s fun to work on it!
I found this to be a very interesting article for one main reason. Travis’s opening comment says he is just doing what he likes to do. I read this to mean he likes being asked questions about BeOS so he can trash on it then explain how his little toy kernel is fun to work on but may never be usefull (i.e. why the hell is OpenBeOS using it?)
Hate to be all negative, but why do we even care what this guy has to say? The whole interview makes him to sound like a cocky smart ass.
But Maybe I am wrong.
Heh heh, this interview just drives home the theory that programmers rarely do it for any purpose other than “to see if I could.” It’s a shame. It’s like the opposite of my problem: lots of ideas and ideals, no skill to program them into reality. Every time I see an interview with a programmer of low-level system stuff I get the impression that they could not care any less if what they were doing was actually useful to anyone. When they get paid to do programming for a set purpose, it’s just a job and pays the bills. Often they are completely passionless about it or downright unhappy with it.
I guess this is another frustrating human condition. Those who have the ability to create, need those who have the skills to guide; those who have the skills to guide need those who have the ability to create.
Otherwise, we’re all just a bunch of boneheads playing with toys…
I think that when the interviewer said “thank you very much”, Travis took that as saying, “thank you for creating NewOS”.
> Travis’s opening comment says he is just doing what he likes to do.
Uh… yeah, kernel development.
And I think that Travis has earned the right to be able to say that he was dissatisfied with working as a kernel dev at BeInc. I surely get the
impression that he’s qualified.
Also, Allen, note that english is probably Emanuele’s 2nd language. I sound
about 2 trying to speak italian — capisce?
Jace -> not exactly. For example: i don’t need someone to guide me. I need motivation. I have some ideas, and i can code. I’m just lazy and don’t have enough motivation to code ;]
Saying “i have ideas but i can’t code. well that’s how the world works: i can guide those with skills.” is just being more lazy than i am ;]
Learning programming language is easy. Sure, if You want code 3D game than You need mathematics too (i’m poor at math, and i don’t like it – i don’t have enough patience and probably to learn it), but not all programming ideas need 3D ;]
> Otherwise, we’re all just a bunch of boneheads playing with toys…
That’s why we do it. It’s a hobby, who cares if we write useless junk? Let us have fun doing it, if we like to.
It’s too bad there’s no way to write up a non-disclosure agreement that
covers interviews like this, I’m sure the OpenBeOS team would empty their
pockets for one. Hope that kernel is better than it sounds, good enough
to compensate for having to deal with this guy.
I think Mr. Geiselbrecht’s attitude is simply realistic. Making conjectures about other people’s motives is always dubious, but I’ll take the risk: I would imagine most of the people who joined Be, Inc. before 2000–and almost certainly all of the ones who were there in 1998 or before–had the goal of making a revolutionary desktop operating system. Problems must have been visible on the inside in 1999, though; early Be newsletters talked about planned upgrades and changes that never happened. In any case, it had to be obvious that the goal wasn’t going to be reached, and it’s natural to think that not all of the people who signed up for that mission would be interested in going after the non-existent “internet appliance” market. (With all due respect to the apparently now-vanished Lefman, the market didn’t exist. Period. Us evil skeptics were right and you were wrong. In your face. Neener neener.)
And, hey, as much as I still miss things in BeOS–just a few minutes ago I had reason to miss, of all things, the “People” files–Mr. G is also right in that it’s nice to not have to worry about hardware compatibility issues.
Very nice post WattsM.
I agree with you.
I know Travis personally (we were housemates for many months) and I know also many ex and Be engineers. What Travis said is exactly what many of these people (including me) believe about the BeOS.
>It’s too bad there’s no way to write up a non-disclosure agreement that
covers interviews like this.
Thanks God there are none. Travis spoke his mind, and for this he is commentable.
>Hope that kernel is better than it sounds, good enough to compensate for having to deal with this guy.
The kernel is better than it sounds and it is indeed a very interesting kernel experiement, with lots of features and techniques not found in any other OS.
Be careful how you talk about engineers over here btw. I will not have people like Travis, Michal or Zac the other day treated with no respect or zealotry.
Travis also mentioned in the interview that he does not know anyone who left Be one way or another and still uses BeOS. So true! Even my husband, only uses Windows these days. I pressed him to install Linux Mandrake in August, he did so, and he deleted that partition 2 days later (he used to use Slackware back in the day and liked it a lot, but he hated the way Linux has become today).
I have no idea why the OpenBeOS project decided to use the NewOS kernel instead of Linux. Linux is well-supported, fast, stable, and has an active developer base. NewOS is, well, new (and not even that new, given that the author doesn’t think he’ll put in any new, crazy, things).
> Allen, note that english is probably Emanuele’s 2nd language
yeah… and not so good
Btw, IMO Travis is not stupid and he wrote (well, he’s still writing it) sth which is experimental but that has a lot of good features. And i think that openbeos’ choice is not so bad!
Ciao a tutti!!
I do apologize if I sounded like an ass on the interview. These net interviews never really come across the way I like them to. You really need to understand the context of the interview, which was to explicitly talk about newos and it’s relationship to Beos and openbeos. Or at least that’s the vibe I got.
Here’s the summary of my point:
1) I work somewhere else now, and don’t use BeOS too much anymore.
2) I spend some of my free time working on a kernel for the fun of it.
3) Some people are using it to build a beos clone. I support them.
That’s pretty much it. I mean no ill will towards anyone, and sorry if I ruffled some feathers.
>I have no idea why the OpenBeOS project decided to use the NewOS kernel instead of Linux. Linux is well-supported, fast, stable, and has an active developer base.
First of all, there is a BeOS clone in development based on Linux and it’s called BlueOS.
Second, while the Linux kernel might be stable and all, I do not think it has things like preemptive multi-threading. Moreover, the multi-processor support, one of BeOS’s strongest features isn’t nearly as good in Linux. Also, the architecture of the BeOS kernel, which after all is the OS they are trying to clone, is totally different from the Linux kernel. Linux basically packs everything into one package, while BeOS has lots of servers running in user land that take care of most stuff (a micro-kernel I think it’s called). I’m no expert, but I believe there are no mature open-source micro-kernels. So I don’t think NewOS is such a bad choise for a BeOS-clone.
Then again, why was it again that they did not use the AtheOS kernel? It seems to me that that one is a lot more mature and pretty BeOS-like. But I’m no engineer/programmer, so what the hell do I know, eh?
In my experience, by the way, Linux isn’t nearly as fast as BeOS. That probably has more to do with that Xwindows-slug-thing than with the kernel, though.
Finally, about Emanuele’s interviewing. I think losing some exclamation marks and smileys here and there, will already help make it look a lot more.. mature.
…while the Linux kernel might be stable and all, I do not think it has things like preemptive multi-threading.
Yes it does, in kernel space by default and user-land through patches.
OTOH I don’t think NewOS has this yet.
Moreover, the multi-processor support, one of BeOS’s strongest features isn’t nearly as good in Linux.
Linux has good SMP support, I don’t imagine NewOS has come very far here… even though it claims to support SMP.
In my experience, by the way, Linux isn’t nearly as fast as BeOS. That probably has more to do with that Xwindows-slug-thing than with the kernel, though.
Well… you answer your own question here.
The problem is that you compare the BeOS kernel to Linux, while we’re talking about NewOS vs Linux. Don’t get me wrong though – personally I think NewOS looks like a better alternative than Linux to base OpenBeOS off, thinking of the projects goals which Linux won’t suite too well.
>>…while the Linux kernel might be stable and all, I do not think it has >>things like preemptive multi-threading.
>Yes it does, in kernel space by default and user-land through patches. OTOH I don’t think NewOS has this yet.
You didn’t look at NewOS that much then. Preemptive kernel with full mutithreading. The kernel thus far is much more reentrant than even BeOS, though you couldn’t tell that unless you had read the source code to BeOS. The VM especially is far more reentrant than the BeOS one, which has quite a few global locks.
>>Moreover, the multi-processor support, one of BeOS’s strongest features isn’t >>nearly as good in Linux.
>Linux has good SMP support, I don’t imagine NewOS has come very far here… even though it claims to support SMP.
Full SMP, with the previously mentioned full kernel reentrancy. Tested on 4 proc boxes. The machine I test it on is a dual proc p3.
I’m not going to say NewOS is the best thing that ever came along. It’s still very immature, but these basic things get taken care of pretty early or not at all.
Please, if you read the feature list on the NewOS web site, take it for what it says. I’m not making stuff up.
I think it was a good interview.
A bit sarcastic, may be.
“Everybody knows that the OpenBeOS project has adopted NewOS as the baseline …” – I really like the “everybody” part.
As one of the answers about why NewOS as OpenBeOS kernel:
I’m not a member of the project (yet), but my impression is that OpenBeOS is trying to recreate the BeOS spirit of pre-IPO days. OpenBeOS newsletter is one example. NewOS kernel is a completely new kernel made from scratch just like BeOS kernel.
Travis, I have a question I haven’t seen yet:
What takes more of your time – improving internals (IDE driver, VM etc) or porting to new architectures ? Or in other words, where your priorities are ?
Sun Blade 100 goes for $700 on e-bay – should I buy it now or better wait for a week 🙂
>Travis, I have a question I haven’t seen yet:
>What takes more of your time – improving internals (IDE driver, VM etc) or porting to new architectures ? Or in other words, where your priorities are ?
It’s much harder to build new functionality, the former. Porting hasn’t been that big of a deal, but I’ve really only spent time porting to one platform, the Dreamcast. The idea was to test out the architecture independance layer and fix some bugs in the design by porting to a weird architecture. It took a few weekends to get it working. Presumably porting to another architecture isn’t that hard, if you dont count all the new drivers you have to write.
>Sun Blade 100 goes for $700 on e-bay – should I buy it now or better wait for a week 🙂
🙂 They’re pretty neat, but if you want it as a unix box, you may be better off with a PC. If you want to hack around on another architecture, it’s a pretty neat box. That’s what I got mine for.
vlad & travis.. can I add this last question and the answer in my interview?