[openbeos] Re: Hi from the PetrOS camp.
- From: "Michael Phipps" <mphipps1@xxxxxxxxxxxxxxxx>
- To: openbeos@xxxxxxxxxxxxx
- Date: Thu, 08 Nov 2001 20:22:05 -0500
Let me respond here, with what I was thinking.
First of all, we have a working kernel. The BeOS R5 kernel.
That is sufficient for 99% of us to develop on.
I asked Peter many of the same questions that you all have.
What do you get out of this? What would you contribute? Why would you help us?
The way I see it, Peter (and co) have been spending years building really cool
stuff.
They want to finish and ship some of that stuff. But they are (basically) where
Be was
in 1993 or so. They have a kernel. They have a networking stack. Nothing else.
Finishing a whole OS from there will chew up *PILES* of cash. Which I would
guess
they probably don't have. Their current product line is either stuff that MS
gives away
(see how well competing with that worked for Netscape) and/or stuff that MS is
targetting as a market (mail servers, www servers). It seems to me that Trumpet
needs a new direction, because with out it, MS will steam roller them. So they
looked
toward OS development. They have a working, interesting kernel. With, of
course, networking.
They need to add the *TONS* of other stuff to it to make it work. You know.
GUI, file system, etc.
And that is a lot of work. More than I would guess that they could afford to
do.
So they are looking to partner.
What do *we* get out of this? Good question. My first question, too.
First of all, we get someone who has actually written a TCP/IP stack. One that
sold a bunch of copies. I am sure that Jean would not turn that down. Oh - and
the help of someone who has built a kernel. I sure won't turn that down. Plus,
we
get known in Australia, and we gain credibility and experience. Peter is well
aware
that we are an open source project (MIT license, BTW). He knows that anything
that
he contributes will be under that license. He also knows that we intend to
finish the
kernel that we have started and give it away.
It has crossed my mind, though, that Trumpet could well be the Red Hat of
Australia for OBOS.
Just a thought.
In any case, Peter, I welcome you. And I think that everyone else will, too,
when they come to understand
the things that we have discussed, as explained above.
>Fair enough. However, I can see the OpenBeos project being split into two
>parts that could operate in parallel to complete the project faster and
>allowing for the BeOS API to be ported to other operating systems. If you know
>of alternative projects which might be more suitable, I'd be interested.
>
>To answer your other question as to what we'd gain, well we'd gain possibly a
>larger user base than we might get if we stuck to Win32 only, plus it would
>demonstrate that our OS is more powerful than being just a Win32 workalike.
>The design philosophy is of being able to support more than one user API.
>
>If the idea works well, your project might gain a little momentum and perhaps
>credibility from a commercial point of view by having a commercial developer
>willing to work with you.
>
>So I guess the answer is "thanks, but no thanks", no?
>
>Peter
>
>On Thu, 8 Nov 2001, Daniel Reinhold wrote:
>
>>
>> Hello Peter!
>>
>> I think it's interesting that you are aware of OpenBeOS and are
>> offering kernel support for it. I'm a bit confused as to what you
>> (Trumpet Software) would get out of such an endeavor. A licensing deal,
>> I presume? I would need to hear more specifics about that before I
>> could really say one way or the other.
>>
>> For myself, I'm perfectly happy with the path we've chosen with using
>> NewOS. Outsourcing the kernel would have some obvious advantages, but
>> it would also have some drawbacks. The idea of designing our own kernel
>> was that the source code for the entire OS from top to bottom would be
>> in our hands and under our control. Unless NewOS has hit some major
>> obstacle that has the kernel team ready to jump ship, I would prefer to
>> continue the track we have already chosen.
>>
>> >
>> >Hi. Just thought I'd introduce myself.
>> >
>> >I have been talking to Michael Phipps about the possibility of
>> building a
>> >BeOS layer to our PetrOS operating system.
>> >
>> >We have a lot of the hard work of networking, OS kernel design and
>> stuff, and
>> >our kernel probably has facilities close to that which a BeOS user
>> >implementation might require.
>> >
>> >While we also do Win32, I've tried to design our kernel so that it's
>> not tied
>> >to any one OS philosophy.
>> >
>> >Anyway, after discussions with Michael, we think we could get some
>> suitable
>> >layer that could graft a beOS subsystem onto it by doing two things.
>> >
>> >1) creating a BeOS kernel driver module to provide BeOS kernel
>> services.
>> >
>> >2) adding an ELF loader as another executable format supported by
>> PetrOS. Our
>> >native format is PE executables.
>> >
>> >Although you are working on your own OS, I estimate that the bulk of
>> your work
>> >is likely to be at the application layer, not the 1/3 as suggested by
>> some. If
>> >we can provide some basic kernel services, it may be a way to
>> acceralte your
>> >development. Also as our networking stack and FS drivers are fairly
>> mature,
>> >you could save yourselves the headaches of reinventing the wheel.
>> >
>> >What is the general opinion and consensus regarding our suggested
>> input into
>> >the project?
>> >
>> >Regards
>> >
>> >Peter
Other related posts:
