The Amiga has a lot of shortcomings, many of which C= was painfully aware of. One of those is applications and application installation.
And before you mention Installer, that is not what I am thinking of: Try to install an application and then move it to another partition or directory. Using WB. Or whatever really - did you get all the dependent files with you when you moved it?
I don't think C= had any definitive solution, but I think they pondered upon having a special kind of directory that contained all belonging files and by naming would know what the executable inside was named.
OTOH, I am still waiting for the *nix world to come up with a proper solution to the nightmare of version dependence. So far they have become very good at doing workarounds, but they behave like they drank metanol for breakfast if you have just a single minor point too low on the release number of a library.
And the viability of the Amiga today?
C= did a telling test with Diskdoctor: They saved it on a floppy disk and then started it, telling it to repair the disk. It basically committed suicide and made the disk useless. After that they simply stopped distributing it.
The analogy would be to start a debugging monitor and write all over memory and see how well the machine survived that (or make a minimal program that does the same). If you go down in flames then you should rethink what you are doing in 2015.
And still, I love the Amiga. I don't see much of a path forwards, but plenty of sideways possibilities. If any of the enhanced FPGA projects ever get released that will be good.
Any path forward wont be Amiga compatible, but it could be _heavily_ inspired (i.e. identical for the end user); there is a bit of research done on SASOS (Single Address Space Operating System - yes, that sounds strangely familiar) which targets 64bit systems. Wait and see. Like always.