[Raion]

I semi quietly had to shelve this project recently, but it was not all for naught. The reasons behind this were a significant amount of the team had to pull out for various personal reasons ranging from medical to psychological. 

I haven't really come clean on it necessarily but suffice to say that it was bad timing. That being said, Endeavor, one of the original team members (originally primarily an advising position) agreed to commit to what he had in terms of time and expertise, and funding was never withdrawn (in fact it was more or less ready to go including the checks already having been signed) 

We've already begun. Libdisk, mount(1) and umount(1) are all critical components of IRIX 6.5.x and are all being reverse engineered. Other targets include the EFS filesystem driver to serve as a blueprint for future driver shenanigans and some parts of IRIX's VFS code. 

We are using a simpler version of the process which with the assistance of LLMs is at least making some progress. But with IRIX as a whole being over 15 million lines of code in the 6.5 releases, and most of what we are touching being under 40k as of this writing... It's unfortunately not scalable so I'm not able to commit to it at length. IRIX 5.3 was around 4 million or so estimated lines, for reference, and being conducted more or less as a trial for larger programs. We did manage to get the full funding from interested private investors including the oceanography company, myself, and two angel investors. Most of that's going straight to Endeavor because he is handling the hard part and I am having to act as the reverse engineering personnel and I am not fantastic at this nor am I the right person to do it at scale. 

At this point we are just tackling areas that we can easily do. Decompiling, documentation and clean reverse engineering and function by function matching all take time. We have to make sure the compile result matches especially for things that are being replaced piecemeal like this. I don't know how much is going to end up being done by me and Endeavor, but suffice to say it'll probably count for less than 5% of the IRC codebase, and be mostly kernel code or simpler userspaec libraries. We are not targeting anything with motif or opengl, very little on the way of platform specific drivers, definitely no graphics code, and most of what we're doing is focused primarily on getting things into usable states, so build flags, build systems and getting everything just right really isn't in the cards.