• Welcome to OS2World OLD-STATIC-BACKUP Forum.
 

News:

This is an old OS2World backup forum for reference only. IT IS READ ONLY!!!

If you need help with OS/2 - eComStation visit http://www.os2world.com/forum

Main Menu

Port of K42?

Started by BigWarpGuy, 2008.01.28, 20:49:05

Previous topic - Next topic

BigWarpGuy

http://domino.research.ibm.com/comm/research_projects.nsf/pages/k42.index.html
Is any one doing or attempting a port of IBM's K42 project to OS/2-eCS?   ???  Since cpu's are now multicore, having an operating system that works with it for OS/2 would be good(?).  ???  8)

Paul Smedley

Tom,

Quote from: BigWarpGuy on 2008.01.28, 20:49:05
http://domino.research.ibm.com/comm/research_projects.nsf/pages/k42.index.html
Is any one doing or attempting a port of IBM's K42 project to OS/2-eCS?   ???  Since cpu's are now multicore, having an operating system that works with it for OS/2 would be good(?).  ???  8)

<Rant On>
I seriously doubt anyone is working on this.  Perhaps instead of posting endless 'Is any one working on xxxx' posts, you could consider spending some time with a 'Learning C' book and a compiler, and learn how to port software.

For reference sake, I did Engineering at University from 1993 - 1996.  Part of that was 6 months studying C.  After that, I dabbled occasionally in C, but not much.  In 2005, I tired of my USB scanner not working properly and learned the usbcalls library and updated Sane to work with modern scanners.

My point is, I was far from being an experienced programmer when I started working on OS/2 ports - yet I've managed to port a bunch of software to OS/2.  If someone like me can do it, many others in the OS/2 community could also do it.
<Rant Off>

Cheers,

Paul.

IBManners

Hi,

> Is any one doing or attempting a port of IBM's K42 project to OS/2-eCS?   Huh
> Since cpu's are now multicore, having an operating system that works with it for
> OS/2 would be good(?).

There isn't much point, its a OS kernel project that would be unrealistic to move
parts of OS/2 onto without access to a lot of resources, including access to IBM's
OS/2 Code. Not to mention adding in to that kernel the interface required for
the WPS and other OS/2 subsections to be able to use.

Be easier to get someone to write a WPS look alike that run's on the Linux kernel
and just use Linux software :-)

Ditto what Paul said to, I'm not a programmer, just a lowly Electronics Engineer who's
worked in Mining/Computer/IT/Com's fields and its very obvious to me that its not
something even worth mentioning. [said nicely ok]

Cheers
Ian Manners
I am the computer, it is me.

warpcafe

Hi,

porting the K42 kernel to OS/2|eCS is much like comparing apples with pears.
K42 is a microkernel. OS/2 is an operating system that uses its own (micro)kernel.

So the bottom line is, that if you want "K42 with eCS", you would have to port the WPS to run on K42, not the other way raound. But then, even if someone manages to overcome the technical obstacles (compiler? DLL? presentation layers...) that gives you a WPS on a kernel and still not an operating systems that is able to do ANYTHING like running whatever programs or accessing files in filesystems... and talking about programs: They will have to be made to run on that Kernel and in that whole execution model.

To tell a long story short:
If we're lucky, we might see Voyager run on K42 in some decades from here.
If you want to get it sooner, spend approx. 12 Million bucks to Netlabs to enable them to hire the required people on a project-dedicated basis.
...or: Give ecosoft 1 Million bucks. They will make a good advertising vor "eCSK42" vaporware on TV, deliver some stuff nobody understands and burn the money for a new launchpad.

Greets
Thomas
"It is not worth an intelligent man's time to be in the majority.
By definition, there are already enough people to do that"
- G.H. Hardy

Criguada

Hi Thomas!

Quote from: warpcafe on 2008.01.29, 11:17:11
Hi,

porting the K42 kernel to OS/2|eCS is much like comparing apples with pears.
K42 is a microkernel. OS/2 is an operating system that uses its own (micro)kernel.

So the bottom line is, that if you want "K42 with eCS", you would have to port the WPS to run on K42, not the other way raound. But then, even if someone manages to overcome the technical obstacles (compiler? DLL? presentation layers...) that gives you a WPS on a kernel and still not an operating systems that is able to do ANYTHING like running whatever programs or accessing files in filesystems... and talking about programs: They will have to be made to run on that Kernel and in that whole execution model.

That's not quite correct. "Porting the WPS to run on K42" is quite a nonsense... the WPS needs a whole set of subsystems to work: in a word, it needs an underlying OS. K42 is a kernel, not an OS. It is something an OS can be based on.
So, if you want to have the WPS running on K42, you'll have to re-implement all the low-level OS/2 APIs based on the K42 kernel. When you have all the low-level APIs in place, you can have all the higher level parts running by just a recompile... if you have the sources obviously. Otherwise you need to achieve binary compatibility by writing (at least) a loader for the LX executable format.
So, before you can have the WPS running on K42, you'll have a running OS on top of that.

The point that we already have an effort to rewrite OS/2 on top of another kernel, and it is osFree. osFree already selected the kernel of choice, and continually debating over this or that "cool" kernel is not helping.

We also already have an effort to rewrite the WPS in such a way that the underlying kernel (or even OS) will not be important. This effort is called Voyager. So, let's support the work in progress.

Quote from: warpcafe on 2008.01.29, 11:17:11
...or: Give ecosoft 1 Million bucks. They will make a good advertising vor "eCSK42" vaporware on TV, deliver some stuff nobody understands and burn the money for a new launchpad.

Greets
Thomas

While I agree that eCOSoft is sometimes taking a questionable approach, I think you're being too harsh on them. After all, a LOT of software lately comes from them, and I would not call it vaporware. Games, WPS enhancements, a lot of things that make eCS look and feel more modern. And yes, a new launchpad ;)
Compare it with the "effort" of *that* german that wanted to SELL... what was it called? PAF? You can see the difference.

Bye
Cris

BigWarpGuy

Quote from: Paul Smedley on 2008.01.29, 04:26:14
Tom,

Quote from: BigWarpGuy on 2008.01.28, 20:49:05
http://domino.research.ibm.com/comm/research_projects.nsf/pages/k42.index.html
Is any one doing or attempting a port of IBM's K42 project to OS/2-eCS?   ???  Since cpu's are now multicore, having an operating system that works with it for OS/2 would be good(?).  ???  8)

<Rant On>
I seriously doubt anyone is working on this.  Perhaps instead of posting endless 'Is any one working on xxxx' posts, you could consider spending some time with a 'Learning C' book and a compiler, and learn how to port software.

For reference sake, I did Engineering at University from 1993 - 1996.  Part of that was 6 months studying C.  After that, I dabbled occasionally in C, but not much.  In 2005, I tired of my USB scanner not working properly and learned the usbcalls library and updated Sane to work with modern scanners.

My point is, I was far from being an experienced programmer when I started working on OS/2 ports - yet I've managed to port a bunch of software to OS/2.  If someone like me can do it, many others in the OS/2 community could also do it.
<Rant Off>

Cheers,

Paul.

I will do my best to learn C/C++ plus other languages.

lpino

Quote from: IBManners on 2008.01.29, 07:56:25
Hi,

> Is any one doing or attempting a port of IBM's K42 project to OS/2-eCS?   Huh
> Since cpu's are now multicore, having an operating system that works with it for
> OS/2 would be good(?).

There isn't much point, its a OS kernel project that would be unrealistic to move
parts of OS/2 onto without access to a lot of resources, including access to IBM's
OS/2 Code. Not to mention adding in to that kernel the interface required for
the WPS and other OS/2 subsections to be able to use.

Be easier to get someone to write a WPS look alike that run's on the Linux kernel
and just use Linux software :-)

Ditto what Paul said to, I'm not a programmer, just a lowly Electronics Engineer who's
worked in Mining/Computer/IT/Com's fields and its very obvious to me that its not
something even worth mentioning. [said nicely ok]

Cheers
Ian Manners

Lots of people believe OS2 is WPS. Well it is not and the option of a shell on top of a UNIX kernel just sucks big time.
There were a lot of comments on response to the IBMs answer and most of them talked about how outdated was OS2 and that Vista and new Linuxs had so many new things. Well what new do they have?, layers to be used on new graphics cards, and support for new hardware, well that-s not much from the OS point of view.

Vista is still the same NT kernel, bloated to fix all the security holes and Linux is just another UNIX kernel, some advance....

Lets just support the ongoing work. Maybe they are right, maybe they are wrong, but al least they are doing something real. Maybe K42 is better than L4 but for now L4 is the one being used.
You have to remember that MacOS X is based on Mach3 a very old microkernel and you have a very massive and solid OS. L4 is a second generation microkernel aimed at reducing the cost of system calls.

Leonardo Pino

warpcafe

Hi Cris,

Quote from: Cristiano Guadagnino on 2008.01.29, 13:01:25
Quote from: warpcafe on 2008.01.29, 11:17:11
(here goes the crap I wrote... ;-)
That's not quite correct. "Porting the WPS to run on K42" is quite a nonsense... the WPS needs a whole set of subsystems to work: in a word, it needs an underlying OS. K42 is a kernel, not an OS. It is something an OS can be based on.
(snip)

I certainly know. And I know that WPS is not OS/2 (while we're at it...).
"Porting the WPS to run on K42" is of course nonsense too (regarding the nonsense "port of K42 to eCS"), the intention was to keep it as short as possible so that BWG understands the principles behind why it simply does not work to "port K42 to eCS"...

So for the big picture from my point of view: ...what we we *could* have is:
"A recode of the WPS from scratch for an operating system that uses a K42 microkernel".
And that certainly requires an appropriate operating system in the first place, huh?
...which implies several layers of architecture (including loads of complicated inter-dependancies that do not exist) around the hardware and kernel, and last but not least a CONCEPT of how the window management works, including graphics ("driver") access, system objects ("model" if you want), with everything based upon the assumption that someone knows how to code and compile the myriads of parts of an operating system on/from a *different* operating system (as the target one doesn't exist so far), based on the the assumption that we overcome possible binary (and other) incompatibilities between dev-platform and the run-target platform regarding compiler and build environment.

And I'm not even talking about device drivers for [whatever-port]-attached [whatever-dumb-crap]-devices from manufacturers that do not only not document their dipshit interface but also most of the time are too damn stupid to adopt generic OEM-strings in files to their own device-IDs... and IF they do, they simply swap parts of their components 3 times a year for saving 0,0001 cent per piece and forget to deliver updated files with it.

And once we have all this up and running, at the end of the day - we have a wonderful whole big piece of NOTHING, since at that point in time we still don't have the most ridiculous little piece of application software for it, including browsers, games, office stuff, picture viewers or even pocket calculators.

Or to use the car example:
Basically it's about building a whole new car from scratch with the only thing supplied is a gearbox, but targeting to make it run on a new kind of fuel and in every kind of terrain.

How does that sound?
:-)
"It is not worth an intelligent man's time to be in the majority.
By definition, there are already enough people to do that"
- G.H. Hardy

BigWarpGuy

My wanting to learn C/C++ was more to do with porting other programs or creating programs for OS/2-eCS. I doubt I will ever be good enough to create an operating system.

All the replies are very informative and appreciated. Thanks.  8)

lpino

Quote from: warpcafe on 2008.01.29, 19:27:59
Hi Cris,

Quote from: Cristiano Guadagnino on 2008.01.29, 13:01:25
Quote from: warpcafe on 2008.01.29, 11:17:11
(here goes the crap I wrote... ;-)
That's not quite correct. "Porting the WPS to run on K42" is quite a nonsense... the WPS needs a whole set of subsystems to work: in a word, it needs an underlying OS. K42 is a kernel, not an OS. It is something an OS can be based on.
(snip)

I certainly know. And I know that WPS is not OS/2 (while we're at it...).
"Porting the WPS to run on K42" is of course nonsense too (regarding the nonsense "port of K42 to eCS"), the intention was to keep it as short as possible so that BWG understands the principles behind why it simply does not work to "port K42 to eCS"...

So for the big picture from my point of view: ...what we we *could* have is:
"A recode of the WPS from scratch for an operating system that uses a K42 microkernel".
And that certainly requires an appropriate operating system in the first place, huh?
...which implies several layers of architecture (including loads of complicated inter-dependancies that do not exist) around the hardware and kernel, and last but not least a CONCEPT of how the window management works, including graphics ("driver") access, system objects ("model" if you want), with everything based upon the assumption that someone knows how to code and compile the myriads of parts of an operating system on/from a *different* operating system (as the target one doesn't exist so far), based on the the assumption that we overcome possible binary (and other) incompatibilities between dev-platform and the run-target platform regarding compiler and build environment.

And I'm not even talking about device drivers for [whatever-port]-attached [whatever-dumb-crap]-devices from manufacturers that do not only not document their dipshit interface but also most of the time are too damn stupid to adopt generic OEM-strings in files to their own device-IDs... and IF they do, they simply swap parts of their components 3 times a year for saving 0,0001 cent per piece and forget to deliver updated files with it.

And once we have all this up and running, at the end of the day - we have a wonderful whole big piece of NOTHING, since at that point in time we still don't have the most ridiculous little piece of application software for it, including browsers, games, office stuff, picture viewers or even pocket calculators.

Or to use the car example:
Basically it's about building a whole new car from scratch with the only thing supplied is a gearbox, but targeting to make it run on a new kind of fuel and in every kind of terrain.

How does that sound?
:-)

Sounds fun...  ;D

chennecke

Quote from: Cristiano Guadagnino on 2008.01.29, 13:01:25
Quote from: warpcafe on 2008.01.29, 11:17:11
...or: Give ecosoft 1 Million bucks. They will make a good advertising vor "eCSK42" vaporware on TV, deliver some stuff nobody understands and burn the money for a new launchpad.

While I agree that eCOSoft is sometimes taking a questionable approach, I think you're being too harsh on them. After all, a LOT of software lately comes from them, and I would not call it vaporware. Games, WPS enhancements, a lot of things that make eCS look and feel more modern. And yes, a new launchpad ;)
Compare it with the "effort" of *that* german that wanted to SELL... what was it called? PAF? You can see the difference.

I rather agree with Thomas. Most software that has come from them is still alpha quality and the usefulness of several packages is rather questionable. Creating eSchemes as a means of consistently managing the look and feel in one place was an excellent idea, mind you, but after what must have been two years now it's still not going anywhere. Things look similar for eSchemes Deluxe. I certainly like the idea of updating the IOprocs/codecs. When I tried the package, the installer was, uh, braindead. Yuri fixed that after I filed a bug report that's been pretty much it development-wise since then. Basically, you can see that reflected in Eugene's behaviour: He's a busy bee always trying to start something but then quickly loses focus. The result is that nothing gets done and many people don't take him seriously any more. His enthusiasm and dedication are commendable but they vanish into thin air like a puff of smoke.

BigWarpGuy

"the intention was to keep it as short as possible so that BWG understands the principles behind why it simply does not work to "port K42 to eCS"... "

It is appreciated that you did keep it as 'short as possible' so that I could understand. I do understand. It is appreciated.

I realize that using 'port' was not exactly accurate. I should have asked if 'anyone was trying to use it as a basis for a new 'OS/2 clone'.

I had asked to see if K42 was worth indicating it as an alternative open source for OS/2 users. Since it is not (from the above posts), I will no longer consider it but concentrate on either OSFree or Voyager to see which one succeeds (like watching HD-DVD and BlueDVD to see which format 'wins').

lpino

Quote from: BigWarpGuy on 2008.01.30, 15:40:08
"the intention was to keep it as short as possible so that BWG understands the principles behind why it simply does not work to "port K42 to eCS"... "

It is appreciated that you did keep it as 'short as possible' so that I could understand. I do understand. It is appreciated.

I realize that using 'port' was not exactly accurate. I should have asked if 'anyone was trying to use it as a basis for a new 'OS/2 clone'.

I had asked to see if K42 was worth indicating it as an alternative open source for OS/2 users. Since it is not (from the above posts), I will no longer consider it but concentrate on either OSFree or Voyager to see which one succeeds (like watching HD-DVD and BlueDVD to see which format 'wins').


OSFree and Voyager don't compete, they complement, so lets hope they both 'win'

BigWarpGuy

With OSFree AND Voyager both winning, OS/2-eCS users win too.  8)

Criguada

Hi Christian!

Quote from: chennecke on 2008.01.30, 15:32:50
I rather agree with Thomas. Most software that has come from them is still alpha quality and the usefulness of several packages is rather questionable. Creating eSchemes as a means of consistently managing the look and feel in one place was an excellent idea, mind you, but after what must have been two years now it's still not going anywhere. Things look similar for eSchemes Deluxe. I certainly like the idea of updating the IOprocs/codecs. When I tried the package, the installer was, uh, braindead. Yuri fixed that after I filed a bug report that's been pretty much it development-wise since then. Basically, you can see that reflected in Eugene's behaviour: He's a busy bee always trying to start something but then quickly loses focus. The result is that nothing gets done and many people don't take him seriously any more. His enthusiasm and dedication are commendable but they vanish into thin air like a puff of smoke.

Christian, do you have experience with the recent versions of their packages? eSchemes v1.46 is a great improvement over the previous versions, and a HUGE improvement with respect to the tools we previously had to modify eCS's look. We badly need this, even if it is only "looks".
PianoPad works well, while eSchemes Deluxe only has some problems with resource leaks.
I understand your feelings towards Eugene's behavior (his "announcements" are sometimes unintentionally humorous), but again saying that what they've done is "a puff of smoke" is not fair and not helping.

Bye
Cris