WindRiver buys BSD - new competition for QNX?

I see on The Register this morning that our friends at WindRiver have
purchased the assets of Berkely Software:
http://www.theregister.co.uk/content/4/18112.html

I think this is their second venture into the UNIX/Linux sphere, and in
some ways, its taking them in the same direction that QNX seems to be
heading. What’s everyones thoughts on where they are going, why, and
what it means for QSSL?


Dale Paus
President
CriSys Limited

(905) 474-9111 ext. 26 voice
(905) 474-0536 fax

“IntelliMap/911: The Only Intelligent Solution”

WindRiver isn’t really new competition for QNX. This serves as somewhat of
a validation of QNX’s direction, since 18 months ago VxWorks wouldn’t even
give you support if you wanted to do xdevel from Linux. It’s gonna take a
very long time, before VxWorks has the same level of Posix integration as
QNX does, so for the most part it’s gonna be some sort of marketing
mechanism (finger in the dyke) for the next few years.

“Dale Paus” <dpaus@crisys.com> wrote in message
news:3ACC7F38.DA4DE6EF@crisys.com

I see on The Register this morning that our friends at WindRiver have
purchased the assets of Berkely Software:
http://www.theregister.co.uk/content/4/18112.html

I think this is their second venture into the UNIX/Linux sphere, and in
some ways, its taking them in the same direction that QNX seems to be
heading. What’s everyones thoughts on where they are going, why, and
what it means for QSSL?


Dale Paus
President
CriSys Limited

(905) 474-9111 ext. 26 voice
(905) 474-0536 fax

“IntelliMap/911: The Only Intelligent Solution”

Previously, Rennie Allen wrote in qdn.public.qnxrtp.advocacy:

WindRiver isn’t really new competition for QNX. This serves as somewhat of
a validation of QNX’s direction, since 18 months ago VxWorks wouldn’t even
give you support if you wanted to do xdevel from Linux. It’s gonna take a
very long time, before VxWorks has the same level of Posix integration as
QNX does, so for the most part it’s gonna be some sort of marketing
mechanism (finger in the dyke) for the next few years.

If WindRiver wanted to add posix support to VxWorks with BSD code, they could
just download the latest copy of {Free,Net}BSD and do it. No need to buy anyone
or anything.

I think it would be more likely that they’re aiming to push BSD into the embedded
space as competition for embedded linux (from the likes of montavista and
fsmlabs) and QNX RTP. Buying BSDi is an easy way of getting the BSD gurus (and
credibility) they need to make it happen.


Cheers - Tony ‘Nicoya’ Mantler :slight_smile:


Tony Mantler | Proud ---- Days since the last
QNX Consulting | of our | 27 |
tony@astra.mb.ca | Record ---- “Gerbil Incident”

“Tony Mantler” <tony@astra.mb.ca> wrote in message
news:Voyager.010406095441.765979A@napkin.astra

Previously, Rennie Allen wrote in qdn.public.qnxrtp.advocacy:
WindRiver isn’t really new competition for QNX. This serves as somewhat
of
a validation of QNX’s direction, since 18 months ago VxWorks wouldn’t
even
give you support if you wanted to do xdevel from Linux. It’s gonna take
a
very long time, before VxWorks has the same level of Posix integration
as
QNX does, so for the most part it’s gonna be some sort of marketing
mechanism (finger in the dyke) for the next few years.

If WindRiver wanted to add posix support to VxWorks with BSD code, they
could
just download the latest copy of {Free,Net}BSD and do it. No need to buy
anyone
or anything.

I’m no lawyer, but I think they would be very concerned about doing this.
IIRC, BSDi is a “clean room” implementation of BSD and thus has purchasable
intellectual property rights (which is what I think WindRiver means by
“business friendly”).

I think it would be more likely that they’re aiming to push BSD into the
embedded
space as competition for embedded linux (from the likes of montavista and
fsmlabs) and QNX RTP. Buying BSDi is an easy way of getting the BSD gurus
(and
credibility) they need to make it happen.

If this were the case it would definately not be competition for QNX since
it wouldn’t be a RTOS. VxWorks kernel with a complete Posix layer would,
however, be more serious competition for QNX. I don’t think that either
WindRiver, or QSSL are terribly concerned about intrusions into their market
space by the likes of montavista et al .

There isn’t much of a business case for having two embedded O/S’s, one
non-rt, and the other rt with radically different API’s. The Bluecat thing
makes only slightly more sense (two embedded O/S’s one non-rt the other rt
with nearly identical API’s); of course QNX makes the most sense 1 RTOS with
a full blown Posix API (QNX had to pay a heavy price to achieve this - 3
generations of OS in the time that VxWorks has had only 1 - but going
forward it gives them a big technical advantage).

“John Doe” <john@csical.com> wrote in message
news:9akt2c$a20$3@inn.qnx.com

There isn’t much of a business case for having two embedded O/S’s, one
non-rt, and the other rt with radically different API’s. The Bluecat
thing
makes only slightly more sense (two embedded O/S’s one non-rt the other rt
with nearly identical API’s); of course QNX makes the most sense 1 RTOS
with
a full blown Posix API

Regarding Bluecat Linux and LynxOS…
Ackording to this page, LynxOS is the only POSIX certified RTOS:

http://standards.ieee.org/regauth/posix/posix2.html

What is the competition LynxOS vs. QNX ?



Mats

Previously, John Doe wrote in qdn.public.qnxrtp.advocacy:

“Tony Mantler” <> tony@astra.mb.ca> > wrote in message
news:> Voyager.010406095441.765979A@napkin.astra> …
[…]
If WindRiver wanted to add posix support to VxWorks with BSD code, they
could
just download the latest copy of {Free,Net}BSD and do it. No need to buy
anyone
or anything.

I’m no lawyer, but I think they would be very concerned about doing this.
IIRC, BSDi is a “clean room” implementation of BSD and thus has purchasable
intellectual property rights (which is what I think WindRiver means by
“business friendly”).

Both FreeBSD and NetBSD are distributed under the (aptly named) BSD liscence,
which lets you do pretty much anything you want with the code, from modifying
and resdistributing it openly, to a wholesale appropriation into a closed
commercial product.

Of course, since BSDi is a commercial product, they couldn’t do the same with
that codebase. Obviously they saw something in BSDi that differentiated them
from a simple tarball of FreeBSD or NetBSD.


I think it would be more likely that they’re aiming to push BSD into the
embedded
space as competition for embedded linux (from the likes of montavista and
fsmlabs) and QNX RTP. Buying BSDi is an easy way of getting the BSD gurus
(and
credibility) they need to make it happen.

If this were the case it would definately not be competition for QNX since
it wouldn’t be a RTOS. VxWorks kernel with a complete Posix layer would,
however, be more serious competition for QNX. I don’t think that either
WindRiver, or QSSL are terribly concerned about intrusions into their market
space by the likes of montavista et al .

I was speaking in the sense of having an embedded OS, realtime or not, with
all the trappings of a modern UNIX system, such as embedded linux (rt or
non-rt), QNX, BeIA (rip), etc.


There isn’t much of a business case for having two embedded O/S’s, one
non-rt, and the other rt with radically different API’s. The Bluecat thing
makes only slightly more sense (two embedded O/S’s one non-rt the other rt
with nearly identical API’s); of course QNX makes the most sense 1 RTOS with
a full blown Posix API (QNX had to pay a heavy price to achieve this - 3
generations of OS in the time that VxWorks has had only 1 - but going
forward it gives them a big technical advantage).

If I were making a shallowly embedded system (webtoaster, kiosk, control
interface), I’d much rather be using a full-featured OS like QNX, Linux or
BSD than a tiny OS like VxWorks.

If WindRiver can’t provide both a full-featured OS and an RT system in one
product, they’re at a disadvantage vs QNX et al. If they can’t provide a
full-featured OS at all, then they’re completely dead in the water when it
comes to shallowly-embedded systems.


Cheers - Tony ‘Nicoya’ Mantler :slight_smile:


Tony Mantler | Proud ---- Days since the last
QNX Consulting | of our | 27 |
tony@astra.mb.ca | Record ---- “Gerbil Incident”

Personally, I don’t get too knotted up about whether an O/S is actually
certified to be Posix compliant, as long as it is.


“John Doe” <john@csical.com> wrote in message
news:9akt2c$a20$3@inn.qnx.com

There isn’t much of a business case for having two embedded O/S’s, one
non-rt, and the other rt with radically different API’s. The Bluecat
thing
makes only slightly more sense (two embedded O/S’s one non-rt the
other rt
with nearly identical API’s); of course QNX makes the most sense 1
RTOS

with

a full blown Posix API

Regarding Bluecat Linux and LynxOS…
Ackording to this page, LynxOS is the only POSIX certified RTOS:

http://standards.ieee.org/regauth/posix/posix2.html

What is the competition LynxOS vs. QNX ?



Mats

LOL, how’d you know that it is, unless it is certified? :wink:
My $20 bet says neither QNX4 nor QNX6 as of now would pass formal POSIX
compliance test if someone attempted.

That said, it appears QNX6 has made serious progress in that area. It is
also much closer to be FIPS compliant than QNX4. So, may be one day they
will be certified indeed.

  • igor

Rennie Allen wrote:

Personally, I don’t get too knotted up about whether an O/S is actually
certified to be Posix compliant, as long as it is.


“John Doe” <> john@csical.com> > wrote in message
news:9akt2c$a20$> 3@inn.qnx.com> …

There isn’t much of a business case for having two embedded O/S’s, one
non-rt, and the other rt with radically different API’s. The Bluecat
thing
makes only slightly more sense (two embedded O/S’s one non-rt the
other rt
with nearly identical API’s); of course QNX makes the most sense 1
RTOS
with
a full blown Posix API

Regarding Bluecat Linux and LynxOS…
Ackording to this page, LynxOS is the only POSIX certified RTOS:

http://standards.ieee.org/regauth/posix/posix2.html

What is the competition LynxOS vs. QNX ?

Mats

Previously, John Doe wrote in qdn.public.qnxrtp.advocacy:

“Tony Mantler” <> tony@astra.mb.ca> > wrote in message
news:> Voyager.010406095441.765979A@napkin.astra> …
[…]
If WindRiver wanted to add posix support to VxWorks with BSD code,
they
could
just download the latest copy of {Free,Net}BSD and do it. No need
to buy
anyone
or anything.

I’m no lawyer, but I think they would be very concerned about doing
this.
IIRC, BSDi is a “clean room” implementation of BSD and thus has
purchasable
intellectual property rights (which is what I think WindRiver means
by
“business friendly”).

Both FreeBSD and NetBSD are distributed under the (aptly named) BSD
liscence,
which lets you do pretty much anything you want with the code, from
modifying
and resdistributing it openly, to a wholesale appropriation into a
closed
commercial product.

Yes, but they wouldn’t own the code base, as they do with BSDi, and as
QSSL does with QNX.


If I were making a shallowly embedded system (webtoaster, kiosk,
control
interface), I’d much rather be using a full-featured OS like QNX,
Linux or
BSD than a tiny OS like VxWorks.

I agree.

If WindRiver can’t provide both a full-featured OS and an RT system in
one
product, they’re at a disadvantage vs QNX et al. If they can’t provide
a
full-featured OS at all, then they’re completely dead in the water
when it
comes to shallowly-embedded systems.

Uh huh, and this is why I think they are heading that direction (Posix
layer over VxWorks).

Posix compliance can be a shifty matter. Take the real time
standard. It is possible to provide compliance without any
additional features. One must merely add the appropriate
…h files that indicate each option is not supported, which
can be all of them.

Mitchell Schoenbrun --------- maschoen@pobox.com

I didn’t say that QNX6 was Posix/FIPS, only that I don’t care whether it
is certified as such. In fact QNX4 is certified as Posix 1 - no FIPS
(I have seen the certificate for it). It is certainly possible for an
OS to be Posix compliant without being certified as such (it must have
been Posix compliant when it was sent to the certifying body, assuming
it passed, so by definition it was Posix compliant before it was sent).

I am more concerned that QSSL, gets the things that most people need
working now (this was my point), eventually the compliance will be to a
level where it could be certified; do I care whether they actually go
out and do it at this point ? no; but I bet marketing does, and they
will do it once it gets there.

The original poster was referring to Posix 2 compliant OSes, which QNX6
may very well be right now, since it now has job control (which was the
main thing stopping Posix 2 certification way back - IIRC).

-----Original Message-----
From: Igor Kovalenko [mailto:Igor.Kovalenko@motorola.com]
Posted At: Monday, April 09, 2001 11:40 AM
Posted To: advocacy
Conversation: WindRiver buys BSD - new competition for QNX?
Subject: Re: WindRiver buys BSD - new competition for QNX?


LOL, how’d you know that it is, unless it is certified? :wink:
My $20 bet says neither QNX4 nor QNX6 as of now would pass formal POSIX
compliance test if someone attempted.

That said, it appears QNX6 has made serious progress in that area. It is
also much closer to be FIPS compliant than QNX4. So, may be one day they
will be certified indeed.

  • igor

Rennie Allen wrote:

Personally, I don’t get too knotted up about whether an O/S is
actually
certified to be Posix compliant, as long as it is.


“John Doe” <> john@csical.com> > wrote in message
news:9akt2c$a20$> 3@inn.qnx.com> …

There isn’t much of a business case for having two embedded O/S’s,
one
non-rt, and the other rt with radically different API’s. The
Bluecat
thing
makes only slightly more sense (two embedded O/S’s one non-rt the
other rt
with nearly identical API’s); of course QNX makes the most sense 1
RTOS
with
a full blown Posix API

Regarding Bluecat Linux and LynxOS…
Ackording to this page, LynxOS is the only POSIX certified RTOS:

http://standards.ieee.org/regauth/posix/posix2.html

What is the competition LynxOS vs. QNX ?

Mats

Previously, Rennie Allen wrote in qdn.public.qnxrtp.advocacy:
[…]

Both FreeBSD and NetBSD are distributed under the (aptly named) BSD liscence,
which lets you do pretty much anything you want with the code, from modifying
and resdistributing it openly, to a wholesale appropriation into a closed
commercial product.

Yes, but they wouldn’t own the code base, as they do with BSDi, and as
QSSL does with QNX.

I fail to see what difference it would make. There are plenty of BSD-derived
UNIXes out there, all wholly owned by their respective companies who charge
fees, set liscencing schemes, etc to their own schedules.

Do a ‘find /usr/include -type f | xargs grep -i bsd’ on your favorite commercial
unix flavour if you don’t belive me. :slight_smile:

Remember, this isn’t GPL-land where everyone has to share and share alike.


Cheers - Tony ‘Nicoya’ Mantler :slight_smile:


Tony Mantler | Proud ---- Days since the last
QNX Consulting | of our | 27 |
tony@astra.mb.ca | Record ---- “Gerbil Incident”

Like I said, I’m not a lawyer, but I suspect there are a lot of issues
that can arise from not owning the source base. I worked in the medical
device business, and I know the FDA has lots of problem with ownership
issues with regards to liability (can you imagine somebody who
contributed some BSD code without compensation being sued when it is
found out that a bug in his code caused the death of a patient ?). The
“no warranty” argument doesn’t fly, because this is exactly what the FDA
does not like (they don’t like it from commercial vendors whom they most
definitely would go after - regardless of disclaimers - in a serious
case of negligence, let alone from someone who never derived any
compensation for his work). My guess is that the only way that the FDA
would condone “ownerless” code being included in a class 3 medical
device is if the original author took ownership, agreed to be held
liable for negligence in developing his code (negligence could be
defined as not following “industry standard” s/w engineering practices),
and placed some sort of performance bond (since he is not a corporate
entity).

That’s the FDA, I imagine the FAA would be even less impressed with the
idea of “ownerless” code in a flight control system (where several
hundred people can be wiped out with a single software failure).
VxWorks is used in both of these applications. LynxOS (which I
suspect has free BSD code in it) has never been approved for use in
these applications (they brag about the fact they are approved for use
in a “supplemental navigation system” - which is a system that may cause
a temporary disruption of flight deck operations if it fails in a non
self-evident manner - e.g. your GPS moving map display fails in a way
that it appears to be working, and shows you on course but you aren’t -
the pilot will realize fairly quickly that the VOR, or inertial nav, or
compass, or visual cues don’t agree with the information presented by
the supplemental nav system, and turn it off)

QNX, being modular, can have all of the “ownerless” stuff stripped out,
in order to qualify as a COTS vendor for these applications. I bet that
with VxWorks architecture, that once they start tying the Posix stuff
into it, it will be nearly inseparable from the rest of the code, and
hence they don’t have the luxury that QNX has of using cheap “ownerless”
code for vanilla commercial applications, and removing it for safety
critical applications; hence, my suspicion that they bought BSDi for
ownership issues.

-----Original Message-----
From: Tony Mantler [mailto:tony@astra.mb.ca]
Posted At: Tuesday, April 10, 2001 8:11 AM
Posted To: advocacy
Conversation: WindRiver buys BSD - new competition for QNX?
Subject: Re: WindRiver buys BSD - new competition for QNX?


Previously, Rennie Allen wrote in qdn.public.qnxrtp.advocacy:
[…]

Both FreeBSD and NetBSD are distributed under the (aptly named) BSD
liscence,
which lets you do pretty much anything you want with the code, from
modifying
and resdistributing it openly, to a wholesale appropriation into a
closed
commercial product.

Yes, but they wouldn’t own the code base, as they do with BSDi, and
as
QSSL does with QNX.

I fail to see what difference it would make. There are plenty of
BSD-derived
UNIXes out there, all wholly owned by their respective companies who
charge
fees, set liscencing schemes, etc to their own schedules.

Do a ‘find /usr/include -type f | xargs grep -i bsd’ on your favorite
commercial
unix flavour if you don’t belive me. :slight_smile:

Remember, this isn’t GPL-land where everyone has to share and share
alike.


Cheers - Tony ‘Nicoya’ Mantler :slight_smile:


Tony Mantler | Proud ---- Days since the last
QNX Consulting | of our | 27 |
tony@astra.mb.ca | Record ---- “Gerbil Incident”

Does “Posix compliant” mean you can ONLY develop Posix complient
applications, or
does it mean you CAN develop Posix complient applications? Purely a
developers
question :wink:

Rennie Allen <RAllen@csical.com> wrote in message
news:D4907B331846D31198090050046F80C903BFE9@exchangecal.hq.csical.com

Personally, I don’t get too knotted up about whether an O/S is actually
certified to be Posix compliant, as long as it is.


“John Doe” <> john@csical.com> > wrote in message
news:9akt2c$a20$> 3@inn.qnx.com> …

There isn’t much of a business case for having two embedded O/S’s, one
non-rt, and the other rt with radically different API’s. The Bluecat
thing
makes only slightly more sense (two embedded O/S’s one non-rt the
other rt
with nearly identical API’s); of course QNX makes the most sense 1
RTOS
with
a full blown Posix API

Regarding Bluecat Linux and LynxOS…
Ackording to this page, LynxOS is the only POSIX certified RTOS:

http://standards.ieee.org/regauth/posix/posix2.html

What is the competition LynxOS vs. QNX ?



Mats
\

Doesn’t QNX have a specific disclaimer that it is not to be used to develop
software
for either life support applications or flight control? Just wondering !

Rennie Allen <RAllen@csical.com> wrote in message
news:D4907B331846D31198090050046F80C903C3DA@exchangecal.hq.csical.com

Like I said, I’m not a lawyer, but I suspect there are a lot of issues
that can arise from not owning the source base. I worked in the medical
device business, and I know the FDA has lots of problem with ownership
issues with regards to liability (can you imagine somebody who
contributed some BSD code without compensation being sued when it is
found out that a bug in his code caused the death of a patient ?). The
“no warranty” argument doesn’t fly, because this is exactly what the FDA
does not like (they don’t like it from commercial vendors whom they most
definitely would go after - regardless of disclaimers - in a serious
case of negligence, let alone from someone who never derived any
compensation for his work). My guess is that the only way that the FDA
would condone “ownerless” code being included in a class 3 medical
device is if the original author took ownership, agreed to be held
liable for negligence in developing his code (negligence could be
defined as not following “industry standard” s/w engineering practices),
and placed some sort of performance bond (since he is not a corporate
entity).

That’s the FDA, I imagine the FAA would be even less impressed with the
idea of “ownerless” code in a flight control system (where several
hundred people can be wiped out with a single software failure).
VxWorks is used in both of these applications. LynxOS (which I
suspect has free BSD code in it) has never been approved for use in
these applications (they brag about the fact they are approved for use
in a “supplemental navigation system” - which is a system that may cause
a temporary disruption of flight deck operations if it fails in a non
self-evident manner - e.g. your GPS moving map display fails in a way
that it appears to be working, and shows you on course but you aren’t -
the pilot will realize fairly quickly that the VOR, or inertial nav, or
compass, or visual cues don’t agree with the information presented by
the supplemental nav system, and turn it off)

QNX, being modular, can have all of the “ownerless” stuff stripped out,
in order to qualify as a COTS vendor for these applications. I bet that
with VxWorks architecture, that once they start tying the Posix stuff
into it, it will be nearly inseparable from the rest of the code, and
hence they don’t have the luxury that QNX has of using cheap “ownerless”
code for vanilla commercial applications, and removing it for safety
critical applications; hence, my suspicion that they bought BSDi for
ownership issues.

-----Original Message-----
From: Tony Mantler [mailto:> tony@astra.mb.ca> ]
Posted At: Tuesday, April 10, 2001 8:11 AM
Posted To: advocacy
Conversation: WindRiver buys BSD - new competition for QNX?
Subject: Re: WindRiver buys BSD - new competition for QNX?


Previously, Rennie Allen wrote in qdn.public.qnxrtp.advocacy:
[…]
Both FreeBSD and NetBSD are distributed under the (aptly named) BSD
liscence,
which lets you do pretty much anything you want with the code, from
modifying
and resdistributing it openly, to a wholesale appropriation into a
closed
commercial product.

Yes, but they wouldn’t own the code base, as they do with BSDi, and
as
QSSL does with QNX.

I fail to see what difference it would make. There are plenty of
BSD-derived
UNIXes out there, all wholly owned by their respective companies who
charge
fees, set liscencing schemes, etc to their own schedules.

Do a ‘find /usr/include -type f | xargs grep -i bsd’ on your favorite
commercial
unix flavour if you don’t belive me. > :slight_smile:

Remember, this isn’t GPL-land where everyone has to share and share
alike.


Cheers - Tony ‘Nicoya’ Mantler > :slight_smile:


Tony Mantler | Proud ---- Days since the last
QNX Consulting | of our | 27 |
tony@astra.mb.ca > | Record ---- “Gerbil Incident”

Ivan Bannon <ivan.bannon@rjginc.com> wrote:

Does “Posix compliant” mean you can ONLY develop Posix complient
applications, or
does it mean you CAN develop Posix complient applications? Purely a
developers
question > :wink:

It means you CAN develop Posix compliant applications.

-David

QNX Training Services
dagibbs@qnx.com

Previously, Ivan Bannon wrote in qdn.public.qnxrtp.advocacy:

Doesn’t QNX have a specific disclaimer that it is not to be used to develop
software
for either life support applications or flight control? Just wondering !

I remember the life support disclaimer. Isn’t flight control necessarily
“life support?” :wink:. This was a long time ago however.



Mitchell Schoenbrun --------- maschoen@pobox.com

“Mitchell Schoenbrun” <maschoen@pobox.com> wrote in message
news:Voyager.010424191633.26809I@schoenbrun.com

Previously, Ivan Bannon wrote in qdn.public.qnxrtp.advocacy:

Doesn’t QNX have a specific disclaimer that it is not to be used to
develop
software
for either life support applications or flight control? Just wondering
!

I remember the life support disclaimer. Isn’t flight control necessarily
“life support?” > :wink:> . This was a long time ago however.

This is really irrelevant. To actually use flight control application you’d
need FAA certification first and I think that’s not going to happen before
QNX (company) becomes SEI-3/ISO-9001 organisation. That’s assuming QNX (OS)
passed POSIX/FIPS sertification which is not the case yet.

With direct life support things get only more complicated I suppose.

  • igor

With direct life support things get only more complicated I suppose.

Actually, the FDA really doesn’t care about Posix/FIPS/ISO for a COTS
vendor. They do care about FDA processes being followed by the actual
device vendor, and that any software that makes up a part of the system
has a “responsible party” (the discussion was originally about using
free software as part of a safety critical system).

Don’t know about FDA, but I’ve heard from involved people that when you
want FAA approval they actually set up office on your premises and keep
walking around, watching and checking. It is like having SEI people on
campus, only worse :wink:

  • igor

Rennie Allen wrote:

With direct life support things get only more complicated I suppose.

Actually, the FDA really doesn’t care about Posix/FIPS/ISO for a COTS
vendor. They do care about FDA processes being followed by the actual
device vendor, and that any software that makes up a part of the system
has a “responsible party” (the discussion was originally about using
free software as part of a safety critical system).