Discussion:
Migrating QNX to Linux
(too old to reply)
TW Burger
2012-02-11 01:10:10 UTC
Permalink
I'm writing an article about migrating QNX based embedded systems and
PlayBook apps to Linux and Android. Has anyone any
experience/ideas/opinions with/on/about this?
Kenbo
2012-02-11 04:27:09 UTC
Permalink
Post by TW Burger
I'm writing an article about migrating QNX based embedded systems and
PlayBook apps to Linux and Android. Has anyone any
experience/ideas/opinions with/on/about this?
Well, you did ask for opinions.... The last two projects I've worked
on have been Linux. I sure miss QNX. With a lack of binary
compatibility, lack of determinism, crappy memory management that
relies on OOM killer thread to kill seemingly random processes,
licensing that turns programmers into lawyers, need to develop a
distribution plan for source code, and a need for on-site kernel
engineers for any serious embedded project I can't understand why
anyone would use it commercially on any project that has real
schedules and limited development resources.

As my last boss, "it's royalty free". Didn't help when nearly everyone
in the company got laid off at least partially due to missed deadlines
and unfixable kernel problems.

Linux kernel engineers do make a lot of money and are in high demand
so I guess it's good for something.

Lest you think I'm totally down on Linux I will say that it is an
excellent development platform.
Kenbo
2012-02-11 05:47:13 UTC
Permalink
Post by Kenbo
Post by TW Burger
I'm writing an article about migrating QNX based embedded systems and
PlayBook apps to Linux and Android. Has anyone any
experience/ideas/opinions with/on/about this?
Well, you did ask for opinions.... The last two projects I've worked
on have been Linux. I sure miss QNX. With a lack of binary
compatibility, lack of determinism, crappy memory management that
relies on OOM killer thread to kill seemingly random processes,
licensing that turns programmers into lawyers, need to develop a
distribution plan for source code, and a need for on-site kernel
engineers for any serious embedded project I can't understand why
anyone would use it commercially on any project that has real
schedules and limited development resources.
As my last boss, "it's royalty free". Didn't help when nearly everyone
in the company got laid off at least partially due to missed deadlines
and unfixable kernel problems.
Linux kernel engineers do make a lot of money and are in high demand
so I guess it's good for something.
Lest you think I'm totally down on Linux I will say that it is an
excellent development platform.
I forgot to mention Linux's brain dead logic of overcommitting memory
when swap is disabled. Or the stupidity of sucking up memory for cache
when it's not needed, wanted or kills performance as it flushes to
slow media when the memory is actually needed for something useful.

Linux is a lot like Microsoft in that way, "We know what you need and
you don't".
TW Burger
2012-02-11 21:57:41 UTC
Permalink
Post by Kenbo
Post by Kenbo
Post by TW Burger
I'm writing an article about migrating QNX based embedded systems and
PlayBook apps to Linux and Android. Has anyone any
experience/ideas/opinions with/on/about this?
Well, you did ask for opinions.... The last two projects I've worked
on have been Linux. I sure miss QNX. With a lack of binary
compatibility, lack of determinism, crappy memory management that
relies on OOM killer thread to kill seemingly random processes,
licensing that turns programmers into lawyers, need to develop a
distribution plan for source code, and a need for on-site kernel
engineers for any serious embedded project I can't understand why
anyone would use it commercially on any project that has real
schedules and limited development resources.
As my last boss, "it's royalty free". Didn't help when nearly everyone
in the company got laid off at least partially due to missed deadlines
and unfixable kernel problems.
Linux kernel engineers do make a lot of money and are in high demand
so I guess it's good for something.
Lest you think I'm totally down on Linux I will say that it is an
excellent development platform.
I forgot to mention Linux's brain dead logic of overcommitting memory
when swap is disabled. Or the stupidity of sucking up memory for cache
when it's not needed, wanted or kills performance as it flushes to
slow media when the memory is actually needed for something useful.
Linux is a lot like Microsoft in that way, "We know what you need and
you don't".
Thanks, these are great points. I like QNX and dislike Linux (and Windows)
for the same reasons. There's also the question of security and Linux's
comparatively giant kernel.
TW Burger
2012-02-12 00:41:10 UTC
Permalink
Post by Kenbo
Post by Kenbo
I'm writing an article about migrating QNX based embedded systems an=
d
Post by Kenbo
Post by Kenbo
PlayBook apps to Linux and Android. Has anyone any
experience/ideas/opinions with/on/about this?
Well, you did ask for opinions.... The last two projects I've worked
on have been Linux. I sure miss QNX. With a lack of binary
compatibility, lack of determinism, crappy memory management that
relies on OOM killer thread to kill seemingly random processes,
licensing that turns programmers into lawyers, need to develop a
distribution plan for source code, and a need for on-site kernel
engineers for any serious embedded project I can't understand why
anyone would use it commercially on any project that has real
schedules and limited development resources.
As my last boss, "it's royalty free". Didn't help when nearly everyon=
e
Post by Kenbo
Post by Kenbo
in the company got laid off at least partially due to missed deadline=
s
Post by Kenbo
Post by Kenbo
and unfixable kernel problems.
Linux kernel engineers do make a lot of money and are in high demand
so I guess it's good for something.
Lest you think I'm totally down on Linux I will say that it is an
excellent development platform.
I forgot to mention Linux's brain dead logic of overcommitting memory
when swap is disabled. Or the stupidity of sucking up memory for cache=
when it's not needed, wanted or kills performance as it flushes to
slow media when the memory is actually needed for something useful.
Linux is a lot like Microsoft in that way, "We know what you need and
you don't".
I prefer QNX too. But ask yourself why are you developing in Linux and n=
ot =

QNX?

There may be a strategic advantage in moving the development platform of=
=

an embedded product from QNX to Linux. Ownership of QNX has gone from th=
e =

original company founded in 1980 to Harmon International in 2004 and to =
=

Research In Motion (RIM) in 2010. Both Harmon and RIM wanted to build =

in-vehicle infotainment systems based on QNX and RIM developed the =

PlayBook tablet based on QNX and seems to have plans to develop QNX base=
d =

smart phones. RIM has recently seen market losses, a loss of prestige su=
ch =

as the Halliburton switch to iPhone, and currently is undergoing some =

corporate turmoil. Also, RIM is primarily a smartphone company that =

developed a proprietary OS, not an operating system marketer or =

development platform vendor.
These events raise the following questions:
=E2=80=A2 Will RIM=E2=80=99s problems jeopardize continued support for Q=
NX?
=E2=80=A2 Will RIM abandon and sell or orphan QNX leaving the future in =
doubt?
=E2=80=A2 Will RIM stop licensing new versions of QNX or renewing older =
agreements =

in order to monopolize QNX technology to its own advantage?
=E2=80=A2 Will RIM be able and willing to provide the services QNX devel=
opers =

require?
Kenbo
2012-02-12 20:43:29 UTC
Permalink
Post by Kenbo
Post by Kenbo
Post by TW Burger
I'm writing an article about migrating QNX based embedded systems and
PlayBook apps to Linux and Android. Has anyone any
experience/ideas/opinions with/on/about this?
Well, you did ask for opinions.... The last two projects I've worked
on have been Linux. I sure miss QNX. With a lack of binary
compatibility, lack of determinism, crappy memory management that
relies on OOM killer thread to kill seemingly random processes,
licensing that turns programmers into lawyers, need to develop a
distribution plan for source code, and a need for on-site kernel
engineers for any serious embedded project I can't understand why
anyone would use it commercially on any project that has real
schedules and limited development resources.
As my last boss, "it's royalty free". Didn't help when nearly everyone
in the company got laid off at least partially due to missed deadlines
and unfixable kernel problems.
Linux kernel engineers do make a lot of money and are in high demand
so I guess it's good for something.
Lest you think I'm totally down on Linux I will say that it is an
excellent development platform.
I forgot to mention Linux's brain dead logic of overcommitting memory
when swap is disabled. Or the stupidity of sucking up memory for cache
when it's not needed, wanted or kills performance as it flushes to
slow media when the memory is actually needed for something useful.
Linux is a lot like Microsoft in that way, "We know what you need and
you don't".
I prefer QNX too. But ask yourself why are you developing in Linux and not
QNX?
There may be a strategic advantage in moving the development platform of
an embedded product from QNX to Linux. Ownership of QNX has gone from the
original company founded in 1980 to Harmon International in 2004 and to
Research In Motion (RIM) in 2010. Both Harmon and RIM wanted to build
in-vehicle infotainment systems based on QNX and RIM developed the
PlayBook tablet based on QNX and seems to have plans to develop QNX based
smart phones. RIM has recently seen market losses, a loss of prestige such
as the Halliburton switch to iPhone, and currently is undergoing some
corporate turmoil. Also, RIM is primarily a smartphone company that
developed a proprietary OS, not an operating system marketer or
development platform vendor.
• Will RIM’s problems jeopardize continued support for QNX?
• Will RIM abandon and sell or orphan QNX leaving the future in doubt?
• Will RIM stop licensing new versions of QNX or renewing older agreements
in order to monopolize QNX technology to its own advantage?
• Will RIM be able and willing to provide the services QNX developers
require?
My opinion is that Linux is becoming popular in commercial products
for reasons that have nothing to do with its technical capabilities.
Management perception is that no royalties means cheap (nothing could
be further from the truth) and that if others are using it, it must be
a smart thing to do. Strategic advantage you say? Seems the strategic
advantage may be to *not* use it rather than to jump on the same
bandwagon as everyone else.

Sure, something could happen to QNX, but if it does it won't have much
bearing on the technical shortcomings of Linux. It would only mean
that companies with the need for a more capable RTOS would look
somewhere else for a solution.
TW Burger
2012-02-13 01:06:27 UTC
Permalink
Post by Kenbo
I'm writing an article about migrating QNX based embedded systems =
and
Post by Kenbo
PlayBook apps to Linux and Android. Has anyone any
experience/ideas/opinions with/on/about this?
Well, you did ask for opinions.... The last two projects I've worke=
d
Post by Kenbo
on have been Linux. I sure miss QNX. With a lack of binary
compatibility, lack of determinism, crappy memory management that
relies on OOM killer thread to kill seemingly random processes,
licensing that turns programmers into lawyers, need to develop a
distribution plan for source code, and a need for on-site kernel
engineers for any serious embedded project I can't understand why
anyone would use it commercially on any project that has real
schedules and limited development resources.
As my last boss, "it's royalty free". Didn't help when nearly every=
one
Post by Kenbo
in the company got laid off at least partially due to missed deadli=
nes
Post by Kenbo
and unfixable kernel problems.
Linux kernel engineers do make a lot of money and are in high deman=
d
Post by Kenbo
so I guess it's good for something.
Lest you think I'm totally down on Linux I will say that it is an
excellent development platform.
I forgot to mention Linux's brain dead logic of overcommitting memor=
y
Post by Kenbo
when swap is disabled. Or the stupidity of sucking up memory for cac=
he
Post by Kenbo
when it's not needed, wanted or kills performance as it flushes to
slow media when the memory is actually needed for something useful.
Linux is a lot like Microsoft in that way, "We know what you need an=
d
Post by Kenbo
you don't".
I prefer QNX too. But ask yourself why are you developing in Linux an=
d =
Post by Kenbo
not
QNX?
There may be a strategic advantage in moving the development platform=
of
Post by Kenbo
an embedded product from QNX to Linux. Ownership of QNX has gone from=
=
Post by Kenbo
the
original company founded in 1980 to Harmon International in 2004 and =
to
Post by Kenbo
Research In Motion (RIM) in 2010. Both Harmon and RIM wanted to build=
in-vehicle infotainment systems based on QNX and RIM developed the
PlayBook tablet based on QNX and seems to have plans to develop QNX =
based
smart phones. RIM has recently seen market losses, a loss of prestige=
=
Post by Kenbo
such
as the Halliburton switch to iPhone, and currently is undergoing some=
corporate turmoil. Also, RIM is primarily a smartphone company that
developed a proprietary OS, not an operating system marketer or
development platform vendor.
=E2=80=A2 Will RIM=E2=80=99s problems jeopardize continued support fo=
r QNX?
Post by Kenbo
=E2=80=A2 Will RIM abandon and sell or orphan QNX leaving the future =
in doubt?
Post by Kenbo
=E2=80=A2 Will RIM stop licensing new versions of QNX or renewing old=
er =
Post by Kenbo
agreements
in order to monopolize QNX technology to its own advantage?
=E2=80=A2 Will RIM be able and willing to provide the services QNX de=
velopers
Post by Kenbo
require?
My opinion is that Linux is becoming popular in commercial products
for reasons that have nothing to do with its technical capabilities.
Management perception is that no royalties means cheap (nothing could
be further from the truth) and that if others are using it, it must be=
a smart thing to do. Strategic advantage you say? Seems the strategic
advantage may be to *not* use it rather than to jump on the same
bandwagon as everyone else.
Sure, something could happen to QNX, but if it does it won't have much=
bearing on the technical shortcomings of Linux. It would only mean
that companies with the need for a more capable RTOS would look
somewhere else for a solution.
You're right. If RIM was not going to follow through with QNX developmen=
t =

I'd buy the company off of them, recommit to hybrid open source licensin=
g =

and might even create an Ubuntu like platform so that you could have a =

phone, tablet, laptop, workstation, and server running QNX and they coul=
d =

all run the same software.
TW Burger
2012-02-17 23:33:32 UTC
Permalink
Post by Kenbo
I'm writing an article about migrating QNX based embedded systems =
and
Post by Kenbo
PlayBook apps to Linux and Android. Has anyone any
experience/ideas/opinions with/on/about this?
Well, you did ask for opinions.... The last two projects I've worke=
d
Post by Kenbo
on have been Linux. I sure miss QNX. With a lack of binary
compatibility, lack of determinism, crappy memory management that
relies on OOM killer thread to kill seemingly random processes,
licensing that turns programmers into lawyers, need to develop a
distribution plan for source code, and a need for on-site kernel
engineers for any serious embedded project I can't understand why
anyone would use it commercially on any project that has real
schedules and limited development resources.
As my last boss, "it's royalty free". Didn't help when nearly every=
one
Post by Kenbo
in the company got laid off at least partially due to missed deadli=
nes
Post by Kenbo
and unfixable kernel problems.
Linux kernel engineers do make a lot of money and are in high deman=
d
Post by Kenbo
so I guess it's good for something.
Lest you think I'm totally down on Linux I will say that it is an
excellent development platform.
I forgot to mention Linux's brain dead logic of overcommitting memor=
y
Post by Kenbo
when swap is disabled. Or the stupidity of sucking up memory for cac=
he
Post by Kenbo
when it's not needed, wanted or kills performance as it flushes to
slow media when the memory is actually needed for something useful.
Linux is a lot like Microsoft in that way, "We know what you need an=
d
Post by Kenbo
you don't".
I prefer QNX too. But ask yourself why are you developing in Linux an=
d =
Post by Kenbo
not
QNX?
There may be a strategic advantage in moving the development platform=
of
Post by Kenbo
an embedded product from QNX to Linux. Ownership of QNX has gone from=
=
Post by Kenbo
the
original company founded in 1980 to Harmon International in 2004 and =
to
Post by Kenbo
Research In Motion (RIM) in 2010. Both Harmon and RIM wanted to build=
in-vehicle infotainment systems based on QNX and RIM developed the
PlayBook tablet based on QNX and seems to have plans to develop QNX =
based
smart phones. RIM has recently seen market losses, a loss of prestige=
=
Post by Kenbo
such
as the Halliburton switch to iPhone, and currently is undergoing some=
corporate turmoil. Also, RIM is primarily a smartphone company that
developed a proprietary OS, not an operating system marketer or
development platform vendor.
=E2=80=A2 Will RIM=E2=80=99s problems jeopardize continued support fo=
r QNX?
Post by Kenbo
=E2=80=A2 Will RIM abandon and sell or orphan QNX leaving the future =
in doubt?
Post by Kenbo
=E2=80=A2 Will RIM stop licensing new versions of QNX or renewing old=
er =
Post by Kenbo
agreements
in order to monopolize QNX technology to its own advantage?
=E2=80=A2 Will RIM be able and willing to provide the services QNX de=
velopers
Post by Kenbo
require?
My opinion is that Linux is becoming popular in commercial products
for reasons that have nothing to do with its technical capabilities.
Management perception is that no royalties means cheap (nothing could
be further from the truth) and that if others are using it, it must be=
a smart thing to do. Strategic advantage you say? Seems the strategic
advantage may be to *not* use it rather than to jump on the same
bandwagon as everyone else.
Sure, something could happen to QNX, but if it does it won't have much=
bearing on the technical shortcomings of Linux. It would only mean
that companies with the need for a more capable RTOS would look
somewhere else for a solution.
It seems that Linux experience for a programmer is a good thing despite =
=

your headaches:

"Eight in 10 (81%) survey respondents say that hiring Linux talent is a =
=

priority in 2012. This pressing matter is particularly evident when Linu=
x =

demand is compared to hiring in other skill sets: 63 percent of hiring =

managers are increasing Linux hires relative to jobs created in other =

skill areas. The issue? A full 85 percent say finding Linux talent is =

somewhat to very difficult, making Linux professionals some of the most =
=

sought talent in 2012."
2012 Linux Jobs Report, Dice and The Linux Foundation February 16, 2012,=

http://www.linuxfoundation.org/sites/main/files/dice_lf_linux_jobs_repor=
t_2012.pdf

Loading...