Haiku OS and its Niche.

[Martin Iturbide]

Today I was reading Haiku OS efforts to find a niche are working.

They are now being included as a "Radio Automation Software" called "Command Center":
http://www.tunetrackersystems.com/about_haiku.html

We can debate if it is a good niche or not, but I found it very interesting to see how an OS (that was in a worst position than OS/2 and that never really got a huge market) is fighting to became adopted.

Today OS/2's niche are the companies that has a "Legacy OS/2 Application" and that to migrate away from OS/2 will cost more than buying eCS. But that niche is only temporary, we need to find new niches for this platform.  But it is hard to compete with other platforms like Windows, that has a huge marketplace, and with Linux and other open source OSes that has the added-value to do not depend on a single vendor.

If today someone invest on a "solution" that runs on eCS people will ask about... What about the OS? Who is developing it? I understand that IBM is not maintaining it...Who has the source code?  ... those all are logical/rational/realistic questions, and are a blocker when you want to create a solution for OS/2 to find a new niche. You don't want to put your business to run on a OS that if something breaks you don't have the source code to fix it and the one who has the source code no longer mains it.

I still insist that open sourcing OS/2 (cloning it) will allows us to do more. It will give us more users that want to experiment with the platform and eventually we will get business solutions to get into niche. If we wait to clone OS/2 after all the "legacy corporate" customers migrate to other platform it will be even harder to accomplish it. Today it is required not to only port more OSS to OS/2-eCS, but also to start thinking on the things that are inside the hood.

When people says it will be impossible to clone OS/2, I ask them to check ReactOS and Haiku OS, they are doing a very interesting progress on their own.

Regards

[Sergey Posokhov]

FireBee could be another example. And AROS too.

[Martin Iturbide]

- MenuetOS - GNU GPL - last update: 07.04.2015  0.99.96
- KolibriOS - ( MenuetOS Fork) GNU GPL - last update: 5 apr 2015
...and notice that I'm just checking non-linux based OSes.

Why those OSes which has less software, less corporate customers and less community organization than this plaform, has the will and the muscle to develop an open source OS and we don't? Where is the millionaire sponsoring them? Did they stole a bank? Why they don't say that it is imposible to make an OS with their hands? Where is the magic? Where is the trick?

[Sergey Posokhov]

Kolibri is a project maintained by Russian developers to keep their CVs up to date. I know them and I know that there's no forces to sponsor them. Moreover, there is no demand in Russia for such software developers at all, because in oil- and gas-driven economy very few people needs software; it is easier to hire two specially trained monkeys. The OS itself is a mess of assembly code with jump operators and global variables, it is obviously a demo project to sell some services (for example, to police or military forces - they would like to harness them, promising "career growth" in the "future"... and maybe, even air conditioner in office space).

[Sergey Posokhov]

Kolibri is a project maintained by Russian developers to keep their CVs up to date. I know them and I know that there's no forces to sponsor them. Moreover, there is no demand in Russia for such software developers at all, because in oil- and gas-driven economy very few people needs software; it is easier to hire two specially trained monkeys. The OS itself is a mess of assembly code with jump operators and global variables, it is obviously a demo project to sell some services (for example, to police or military forces - they would like to harness them, promising "career growth" in the "future"... and maybe, even air conditioner in the office).

[Olafur Gunnlaugsson]

FireBee could be another example. And AROS too.

FireBee is hardware, the OS is a separate project

......

I own 3x Atari Falcons and have been mulling getting a FireBee for the last couple of years.

[Olafur Gunnlaugsson]

- MenuetOS - GNU GPL - last update: 07.04.2015  0.99.96
- KolibriOS - ( MenuetOS Fork) GNU GPL - last update: 5 apr 2015
...and notice that I'm just checking non-linux based OSes.

Why those OSes which has less software, less corporate customers and less community organization than this plaform, has the will and the muscle to develop an open source OS and we don't? Where is the millionaire sponsoring them? Did they stole a bank? Why they don't say that it is imposible to make an OS with their hands? Where is the magic? Where is the trick?

You obviously have never tried to use any of them in anger .... They are in no way usable on a day to day basis

Haiku committed hara-kiri in the naughties when the developers of it went on a rampage against ZetaOS, not realising that by doing so they eroded the BeOs market in general and lost the only BeOs developer that was actually *paying* people to write open source software.

It took the Haiku project over 10 years to go from start to a full release, by then most of the software they supported was 15 years old and the OS design they copied down to a tee was 20 years old and outdated, and almost all interest in it had dried up. If you start a similar OS/2 project today, by the time you finish up with anything useful, you will have a piece of OS that runs 30 year old software and a 40 year old OS design.

IBM intended to have replaced the original design of OS/2 by 1997 with a more up to date and more coherent variant, rather than the rather thrown together OS they had on their hands, replicating the original design in 2025 is such a strange idea........

[Fahrvenugen]

A few things to keep in mind.

TuneTrakker has been available for quite some time, originally it was developed to be a radio automation system to run on top of BeOS.  When BeOS development stopped, from what I understand they had an agreement to allow them to continue to distribute their own distribution of BeOS with their product (similar to how OS/2 is distributed as a part of eCS).  I believe for a while they also distributed their system on top of Zeta too.  Then eventually to Haiku once it developed enough for actual use. 

The only other (business) options that Tunetrakker had would have been to either try and port to another OS (Windows or Linux most likely), which likely would have required a significant rewrite.  Or to just close up new sales and go into maintenance mode (maintaining current customers). Once Haiku reached the point of being a solid OS, it was the obvious route to go from a business standpoint.

I wouldn't say this is a "niche" for Haiku, since quality radio automation software does exist for Windows, Linux, and Mac (and yes, I do know this as I work in the radio broadcast industry).  Bundling Haiku with the TuneTrakker / Command Center product seems more like the people behind TuneTrakker using the opportunity that Haiku offers for them to be able to continue to offer their product and keep their business moving forward.  Of course there is nothing wrong with this, and it does provide a path for TuneTrakker to continue to be updated and improved, providing upgrade and support options for both current and potential new customers of the software.

You might not know this, but there was at one point a radio automation system developed to run on top of OS/2.  It was called Startrax.  While it was a solid system for its day, I'm not aware of any stations currently running it any more.

IBM intended to have replaced the original design of OS/2 by 1997 with a more up to date and more coherent variant, rather than the rather thrown together OS they had on their hands, replicating the original design in 2025 is such a strange idea........

This is pretty much what the original plan for the PPC version was.  Keep in mind - the PPC version of OS/2 wasn't just a port of OS/2, it was a new OS framework based on top of a microkernel.  The idea was that you'd have a microkernel for a particular hardware architecture, and then run the OS environment of your choice on top of it.  The plan was to have multiple OS environments available, and have microkernels for different architectures.  OS/2 happened to be the only OS environment that reached any point of near completion when that whole project got cancelled.  Some of the remains ended up in Warp 4, Warp Server, and some eventually into Warp 3 via fixpaks, but most of what's left of that project - at least what we've seen in public - is on the OS/2 for the PPC beta CDs that have turned up from time to time.

[Martin Iturbide]

Hi

Take notice that Linux was a clone of Unix, and that the first versions of linux (GNU Linux, the whole package) sucked so bad as the other open OSes that we are discussing. The first version of Linux was unstable and not recommeded for serious works (and should no be used in anger too ), later it was told it was a OS for geeks, that it was so compliated and that lacked of software, now a version of Linux is running in each Android phone. The same architecture of the old Unix (1996) and Linux kernel (1991), with some tweaks, is running today. Maybe Linux has evolved from Unix, but the key was to be open source and not depending on a single vendor.

So, to be it is better to have a OS that sucks that is open source, that an OS that is good but the author no longers developes it. Why? Because there is a forth dimension called "time". By being open source an crappy software can be converted into something stable. By being close source when the developer/company runs out of money or will, the software will be stored in a drawer without any possibility to evolve.

I was told by some guy "Good Software gives you a User Community, bad open source software gives you a Developer Community".

Since we are in the OS/2 community, and we are want to have a future with the same experience we have using this platform (I guess that is the idea), I promote the idea of cloning OS/2 as open source. I know it is now as simple as it sounds, maybe it will be simpler to switch to an OSS OS, but that will be to easy and without fun, right?

So, do not worry if it an open OS/2 will suck at first, that is part of the evolution of open source software. But we have an advantage over HaikuOS and others. We still have eCS that is being sold so we can try to replace the close source components little by little on eComStation until we get a full replacement. We don't need one "big bang" change, we can go giving little steps to replace PM, SOM, CPI, etc... file by file if it is necessary.  Or if someone wants a "Big bang" change with open source and OS/2 we can also give it try

But what I need right now is users and developers with the interest of talking a look under the hood and see what can be replaced.

Regards

[Andi B.]

We don't need one "big bang" change, we can go giving little steps to replace PM, SOM, CPI, etc... file by file if it is necessary.

That is a dramatic wrong wording IMHO. It does not need 'little' steps but giant leaps to replace only one of these parts with open source alternatives that have nearly the quality level we are used to and which OS/2 - eCS users are expect before the even want to try out your replacement component.

You will not find the community which is willing to test and bug fix any OS/2 replacement component when it sucks like the first Linux versions or the examples you've stated above. None of the serious OS/2 developers will only waste a minute looking into messy code as Sergey describes above. It is a big difference between tinkering with some code snippets and coding medium sized quality level applications (read usable replacement parts).

Most probably you will not

...need right now is someone with the interest of talking a look under the hood...

as this was done hundred of times before. The conclusion was always the same - it would need way to huge investment and man power to only replace one of them for our community. Even when we were 10 times as much and 10 times as rich as we are your dreams are so much out of reach you probably can't even imagine. Sorry I've to say that

Think about why only very few people even test the os4 kernel. This is an example of a piece of software which is not that bad at all. In fact it works remarkable well on a lot a machines. But beside the legal concerns which I don't want to discuss here it has a few limitations. This few limitations (documentation, trace support, incompatible with ACPI) prevents usage and testing by a wider audience. Why do you think a lot of people want to play with replacement components which for sure will have a lot more bugs and incompatibilities when you do not even find the testers for much more advanced projects? Why do you think a broader audience is willing and qualified to develop, document, make install packages and test a replacement component when we do not even have the man power to support running projects like AOO, Firefox, Seamonkey, Apache, all the ported QT apps..... ?

[Martin Iturbide]

Why do you think a lot of people want to play with replacement components which for sure will have a lot more bugs and incompatibilities when you do not even find the testers for much more advanced projects? Why do you think a broader audience is willing and qualified to develop, document, make install packages and test a replacement component when we do not even have the man power to support running projects like AOO, Firefox, Seamonkey, Apache, all the ported QT apps..... ?

Those are not questions, only excuses to not try it. "Reality Check" is new wording for "Defeated without trying".  We need a strategy for the future of this platform, the platform does not attract new users because we don't show a future or an attement to have a future.

We always discuss it and always see this two ways of thinking::
1) Think about a future of the platform.
2) Remain as a legacy and live of the good memories, and just take whatever it comes.
(Maybe there are other in the middle)

I prefer the first one. The ones that prefer the second, just move on to the next thread, nobody is judging.

For the moment I'm just in search for the people that want to have a future with this platform. I don't want to find people that wants to code it all, I just want someone that wants to make little steps like the ones I'm doing on OS2World and on EDM/2.

Regards.

[Andreas Schnellbacher]

Andi, I confirm everything you wrote.

Think about why only very few people even test the os4 kernel. This is an example of a piece of software which is not that bad at all. In fact it works remarkable well on a lot a machines. But beside the legal concerns which I don't want to discuss here it has a few limitations. This few limitations (documentation, trace support, incompatible with ACPI) prevents usage and testing by a wider audience.

I have noted down that the ACPI incompatility was fixed. The info came from a posting in this forum. Am I wrong?

(I haven't tested that myself, because my new machine still lacks stable network connectivity.)

[Sergey Posokhov]

Both "ACPI.PSD" and "OS4APIC.PSD" works with OS/4 kernel, both debug and release versions of the kernel.

I also had some troubles with network connectivity: my local service provider ("SmileNet" a.k.a. "InfoLine") uses too small MRU packets... the only SafeFire can work with it, using 1480 byte MRU.

[Mike La Martina]

While I have never seen Kolibri code and cannot offer comment on it, I had an entire career as an Assembly Language programmer, at IBM and elsewhere, and I am amused about Sergey's comment about jumps and global variables.
I have sold millions of US dollars worth of software, world wide that was (is) 100% assembly language.  Neither jumps nor global variables are evil constructs if managed properly.  I can offer no specific opinion except about code that I have personally written.
Since I am now retired and have no more interest in programming, I wish no argument.  FYI I am the author of a withdrawn OS/2 application called Naviplex.  I did write that in C++.

[Dave Yeo]

Both "ACPI.PSD" and "OS4APIC.PSD" works with OS/4 kernel, both debug and release versions of the kernel.

I also had some troubles with network connectivity: my local service provider ("SmileNet" a.k.a. "InfoLine") uses too small MRU packets... the only SafeFire can work with it, using 1480 byte MRU.

Have you (and others as well) changed the line mtudiscover to 1 under the discover column in %ETC%INETCFG.INI ? For some stupid reason eCS turns this off (0) which breaks some systems/routers.