Does anyone know, how many QNX inspired operating systems are there?.
Commercial or otherwise.
I think it would be very hard to say who inspired who. But i’d say none.
What about RadiOS. Though I do not get the point of writing the kernel etc
entirely in assembly. And BTW, what would be QSS’s reaction if someone
wrote a OS somewhat similar to QNX6 in architecture, and made it
commercial.
I think QSS patented some stuff, the message passing. Why you’d want to write the kernel in assembly: get every little bit of speed, some stuff is hard to do in C like playing with pages, setting up LDT/GDT etc, on x86 not all ring 0 instructions are available in C.
Sorry Mario. But about a OS ENTIRELY in assembly like RadiOS, if the point
is to make it public so that others can learn from it, the learner should be
a fanatic of assembly. Otherwise forget RadiOS for say 2 months, come back,
one would be hard put to remember the point one left off earlier.
And to drag on QSS's patenting of message passing etc, what if
someone created components like Process Manager and Send/Receive/Reply
similar in overall idea to QNX Neutrino, plus some additions, in their own
microkernel OS, but do not know how these components actually work in QNX
Neutrino, would that person find QSS lawyers knocking at his door at
midnight.
I guess if they patented it, it’s not to protect themselves against code theft… I don’t know what are the detailed of that patent though.
A kernel is not a big thing in itself, but it’s all the stuff around it that is really lots of work. So just writing a kernel and making it commercial would not really compete with QNX IMHO, even if the kernel is similar feature wise. When you think about it QNX kernel has a very limited feature set.
I went to the US patent web site but couldn’t find it. It’s worth going to have a look though, this patenting stuff is really getting out of hand…
UPDATE
I went to the Canadian Patent site (QSS is after all a Canadian company). And I found a bunch of patent from QNX
patents1.ic.gc.ca/details?patent … anguage=EN
patents1.ic.gc.ca/details?patent … anguage=EN
patents1.ic.gc.ca/details?patent … anguage=EN
patents1.ic.gc.ca/details?patent … anguage=EN ( that is the one about message passing I think, seems kind of weak to me.)
patents1.ic.gc.ca/details?patent … anguage=EN
Mario, thanks for the patents-search. I agree with You that patenting ideas
on transparently distributed message-passing is too vague. Especially when the claims on this topic by QSS goes like this :
patents1.ic.gc.ca/claims?patent_ … anguage=EN
And about the asynchronous messaging :
patents1.ic.gc.ca/claims?patent_ … anguage=EN
are they talking about Neutrino Pulses, or UNIX-like Signals. If the
latter, then Signals have been there for ages now.
From reading all those patent-claims, I could only grant that the one
about Photon is the one which I think they can fairly claim as their own.
So would all this mean a non-start for those who want to or have written
microkernel + QNX Neutrino-like kernels (or OSs) and want to make their
money from it.
Signals dont have the concept of channels, also signals dont carry payload (messages). Signals are not queued. Quite different the async messages.
I really doubt you would have a problem from QNX or anyone else trying to sell a microkernel. You might have a problem if you infringe on their patents.
Ofcourse, silly me.
Ah. That puts, as is said in India, life into my life. But does anybody know
the experiences of the writer of RadiOS (which is similar to QNX Neutrino)
when he might have perhaps tried to sell solutions built around RadiOS.
But QSS could also be flattered if people start creating what are more or
less, clones of QNX.
I am a recent convert (since 5 years) to microkernels. So maybe
I am too zealous in propagating this architecture, by word and deed. But
what has been the experience of the oldies (heh heh) of microkernel.
I can’t speak for QSS, but unless your a big startup company with 50 employes or more I wouldn’t worry about QSS comming down on you. There are plently of company out there that sell microkernel , but they are in a different market segment then QSS. Although the kernel might be an important part of why you’d pick the OS, it really more about the gravy around the kernel that separate them.
My OS (Dragunovos) cannot claim that now atleast its gravy is tastier than in
QNX. There’s no GUI and multiprocessing. So no support for self-hosted IDE and
desktop or server application. But all this will come about in 8 to 9 months.
But even now I say to people that QNX is the best OS in the world. As I have
read in these pages earlier : ‘Use Linux if You want to run a file-server, use QNX if
You want to run a nuclear reactor’. Me thinks QSS is flattered with all this
attention. But I ask again to those who have been using QNX for a long time,
are they still carrying the flag for microkernels, or have they been disillusioned?.
This sounds like a religion question. Excitement and disillusionment can come at many levels.
I never really cared whit about microkernels per say. I do like message passing, that it is at a low level in the OS, and that it bridges the network almost transparently. I like real time. I like reliability. I like writing drivers outside the OS. If I’m told that this is because of a microkernel architecture, ok I like microkernels. I don’t particularly like the fact that I’ve had to learn QNX 3 different times. I don’t like the fact that there is very little contract work for QNX anymore. The fact that this is related to a marketing decision on QSSL’s part, well I understand that they needed to do this for their survival.
Can Linux provide me all I need in an OS now? Probably. Does that make me disillusioned with microkernels? Does that make me give up on QNX? No?
Anyone that thinks one solution is a solution to everything need to go out and smell the world. Microkernel has advantages and disavantages. I’m sure there are better micro-kernel out there then the QNX micro-kernel. Personnaly the advantage I see is no need to write code in kernel mode, as for the rest.
A micro-kernel is so “simple” that it’s very hard to sell a product like QNX based solely on the micro-kernel. And think about it, the more micro the micro-kernel is the less it has to do. I mean an kernel like QNX isn’t very difficult to do feature wise.
If you look on the web you’ll see lots of people having writing their own kernel, it’s not that much work and it’s not rocket science. But the trick is what comes around the kernel, which is what justify the amount of people working at QSS
Mario,
I’m not saying it is the kernel by itself. Clearly a kernel without the stuff around it is barely an OS. This issue, for me at least, really is architecture. While it may not be rocket science to build a kernel, it probably is to know what to put in it. Dan and Gorden have probably had at least 3 opportunities to test this out. I’m not trying to compare QNX to the type of kernel you would put together yourself either, but rather when you put it next to monoliths like XP, and Linux. JMHO.
Mitchell,
I was responding to LeMercenaire last comment
Fine Mitchell, You like microkernels. But You also say…
For anyone who has changed the node-name on a Linux machine and had
to reboot, he must have wondered why did he needed to do it. But I agree
when You say…
Indeed. Small things in QNX like reception of signal by resource manager
when the client unblocks, or running a ISR in the context of its process…
this all is architecture and good one too.
Yes Mario, there indeed may be better microkernels than QNX. I have
myself found some during web-research. Though QNX seems to have
somehow caught attention of people, and one has to doff their hat to QSS
to start-off with a microkernel when the world was going monolithic.
But all said and done, I am not a QNX-fanatic. I have long since woken
up and smelt tea. Having looked at QNX should not ofcourse stop us from
looking at better microkernels, or say the next architecture in computing.
Not The Indian Software Engineer,
LeMercenaire.