AROS has made a deliberate architectural decision not to implement Unix-style shared object support (

dlopen()

,

dlsym()

, etc.) as part of its native operating system model. This was not an oversight, nor is it simply a missing feature waiting to be implemented. It was a conscious design decision based on the architecture and philosophy of AmigaOS. AmigaOS already has its own shared library mechanism, with its own conventions, interfaces, lifecycle management, and integration into the operating system. Unix-style shared objects follow a fundamentally different model that does not naturally fit within the AmigaOS architecture. AROS is an AmigaOS-compatible operating system, not a Unix compatibility layer. We recognise that the absence of Unix-style shared objects can make porting certain software more difficult. However, ease of porting is not, by itself, sufficient justification for reversing established architectural decisions. Porting software to another operating system means adapting that software to the target platform. It does not mean requiring the target operating system to abandon its design principles to accommodate assumptions made by software written for an entirely different environment. Introducing a second, fundamentally incompatible shared library model simply to make Unix software easier to port would add complexity, increase maintenance costs, and potentially undermine the existing AmigaOS APIs.

I think that's reasonable. Not every compatibility API needs to become part of the native OS API. But out of curiosity, doesn't AmigaOS 4 support dynamically loaded

.so

libraries? AmigaOS 4 Libraries I'm wondering whether there's room for an officially supported, optional compatibility layer built on top of native AROS mechanisms, without changing the underlying architecture. Or should such functionality always remain entirely outside the AROS project?

AROS isn't based on OS4 (which wasn't an "official" commodore OS) - and doesn't follow decisions made in it.

Sure, I'm not suggesting that AROS should follow decisions made in AmigaOS 4. My point was simply that AmigaOS 4 shows how the original Amiga architecture can be extended without abandoning its foundations. But point taken.

Im not sure I would say running .so's ontop of AmigaOS is "extending the architecture" - it doesn't integrate with any of the amigaos side. I also dont see that it added particular value to amigaos. Its just a stepping stone to using Linux, which is what they now push on the users of the hardware they sold OS4 to.

That's actually a risk I see and can agree with. I don't think AROS should become Linux bis.

AROS would largely become pointless at that time - you may aswell not run it at all