Petition To Get OS/2 Open Source

Landreth writes "There is currently an ongoing petition taking place at OS2 World to get IBM to open source either the whole part or parts of OS/2 to the community. I would highly encourage the Linux community to take part of this open source petition as well due to the fact there are lots of interesting code base the they could benefit from. To sign the petition: http://www.os2world.com/petition/" Despite the jokes about it, there was some good stuff in OS/2; however, I'd rank the ability to open it up fairly low, since I suspect there's a fair amount of legal restrictions on elements of the code.

Comments

It's a shame

[brennanw]

That even *some* of the code -- specifically the workplace shell -- can't be released as open source. The workplace shell was one of the most elegant and powerful user interfaces I've ever worked with. It wasn't always the most *attractive* interface -- not by default, at any rate -- but it was the only one I've ever used that ever "felt right" to me. I miss that. The phrase "drag and drop" simply didn't do it justice.

Anyway, I signed, but I'm afraid that 1) there's too much proprietary licensed code for the entire thing to be released, and 2) IBM has neither the patience nor the interest in doing the work necessary to separate what can be released from what can't be released. Which is a pity.

Re:MS Code?

[His name cannot be s]

Maybe I'm way out in left field, but wouldn't open sourcing OS/2 open what would likely be a lot of Microsoft's NT code? Weren't OS/2 and NT once the same operating system? I wouldn't be surprised if there were still a bit of shared codebase.

Not quite.

OS/2 3.0 NT was supposed to be built from NT's codebase. Obviously, that didn't happen, and microsoft took their toys and went home and made Windows 3.1 NT.

There is significant evidence that NT 3.1 (and later) Windows 32-bit APIs were influenced by OS/2 's design. The WinScrollWindow api under OS/2 has exactly the same signature as ScrollWindowEx under Win32... The win16 api does not quite match. There are a large number of these close matches in the Win32 API :: OS/2 API.

That being said, NT (and its derivatives) do not share code with OS/2 in implementation. (other than code that was inherited from OS/2 1.3 (ie: HPFS).

It'll never happen

[LouCifer]

And here's why.

IBM sold OS/2 off and it became eComStation ("jointly developed" - whatever). I highly doubt big blue has exclusive rights to the code anymore.

Go ahead and sign the petition, we all know how much weight internet petitions carry.

I, for one, would love to see both of these pan out. Unfortunately they probably won't.

Re:OS/2 Ahh the memories

[maxwell demon]

Well, according to their web page, the ReactOS people actually plan an OS/2 subsystem. Therefore if IBM released the part of the OS/2 code which they can, it would probably be a big help.

BTW, if the OS/2 kernel code is too encumbered, even releasing the WPS alone could be a great thing. While it certainly lacked some features which modern desktops have, it had some other features which AFAIK are still not available on other systems (e.g. what was called "Arbeitsordner" in the German version; essentially a folder which managed its own "sub-session").

Re:Dare to dream!

[LordNimon]

I used to work at IBM on OS/2, and I've written a bunch of code for it. I can tell you that the ONLY code of value that isn't implemented somewhere else is the WorkPlace Shell (WPS). Technically, the WPS could probably be released as open-source. However, that wouldn't help a whole lot, for three reasons:

• The WPS API is well documented and stable. It also isn't that big, since it's just a core API. The real value is in the plug-ins, and there are already plenty of open-source third-party plug-ins for the WPS. It would not be that difficult for someone to recreate the core WPS from scratch, just from looking at the API.
• IBM's code is heavily tied to the Presentation Manager (PM), which is the OS/2 equivalent of the X Window API. You'd spend as much effort trying to rip out the PM dependencies as you would just rewriting the damn thing from scratch.
• As powerful as the core WPS is, the current implementation is pretty weak compared to what KDE and GNOME do today. Even after porting the core WPS to Linux (or whatever), the developers would then have a lot of catching-up to do to make it more desireable than KDE or GNOME.

Perhaps some of OS/2

[AaronW]

Having developed device drivers for OS/2, I doubt there'd be that much interest in the OS/2 kernel or device drivers. Even in Warp, and OS/2 4.0, most of the device drivers were 16 bit since the device driver API was only 16 bit (except graphics drivers). I think maybe the only interesting parts would be the Workplace shell and SOM, though I wonder about the stability in today's complex environment, having remembered having issues of stability with the WPS when I loaded up all the software I ran.

There's also still a lot of Microsoft bits and pieces of code in there.

-Aaron

Re:I'm In (293) - Many More Needed.

[jemfinch]

That means that you cannot compile GPL(or LGPL) code and CPL code together

No, that means you can't compile GPLed code with CPLed code and distribute the resulting binary.

Repeat after me: copyright affects distribution, not use.

Jeremy