>>> There's this ongoing myth that you can just recompile code
>>> from one ISA to another.

>> That's not a myth but a fact, assuming the code is in a high-level language,
>> there's a compatible compiler for both ISAs and the ISAs can operate in the
>> same endianness, which is true for m68k and PPC.

> No, that is simply not true. I've ported software across platforms with the
> same ISA and had to overcome all sorts of architectural issues. Across
> ISAs on hardware that bears no resemblance to the original platform?

You are right. I forgot to include the assumption that the code to be recompiled must not be specific to (aka hardware-banging) components that are missing on the target hardware. You might remember that with the help of CGX and AHI, OS3.x runs on the Draco which lacks the Amiga chipset. Hyperion licensed P96 and AHI for OS4, both of which existed long before work on OS4 began, so they didn't have to invent anything new in this regard. Porting P96 to PPC was probably little more than just a recompile, and AHI was already successfully ported to PPC (PowerUP) before.

> Endian issues and available compatible C compilers are not an issue,
> but the fact that OS3.1 was not completely coded in C is.

Hyperion didn't use Commodore's ASM-ridden code but Olaf "Olsen" Barthel's cleaned-up and ported-to-C code that also is the basis of OS3.5, OS3.9 and OS3.1.4. Of course, some small low-level components remained in m68k ASM and had to be rewritten in PPC ASM for OS4, but this was expected. Problem was that Hyperion planned to somehow built OS4's low-level components upon WarpOS source code which turned out to be barely commented ASM hell and thus was scrapped, so had to be redone from scratch.
In summary, I share takemehomegrandma's opinion that regarding the original plan as outlined in the 2001 agreement, the targeted 6-month timeframe was not an unrealistic one.