I say the CAS OS and Non CAS OS are identical they are just configured slightly differently look at the file sizes of the OSes anyone notice that the NON CAS OS is bigger despite having a less complicated task a CAS is big an i doubt its smaller then the 84 emulator packaged with the Non CAS Nspires OS case and point the non cas nspire is the same os only crippled all we have to do is figure out how to un-cripple it which likely comes down to either a  configuration file change or the modification of a few pointer and we have everything.  If i had an unencrypted version of both OSes I'd find the parser and find the array of function pointers that point to the code executing the commands and start there.  But that's just me.

Wow that makes me even more confident that they are just code switches

that said, I wonder if those commands code is really in the non-CAS os, though. They most likely used most space for the 84+ emulator instead. Is the CAS OS considerably larger in file size than the non-CAS one?

I personally would prefer to not host copies of the OSes (or tools made purposely to install a CAS OS on a normal Nspire) myself on the site if they ever came out, though, after what happened on TI-BANK with the other OSes. I may tolerate links to them, but not hosting them directly, in case TI would send a DMCA notice directly to 1and1 instead of myself. In links case, 1and1 wouldn't be able to do much besides maybe asking me to remove the links, because I wouldn't be hosting any illegal files.

Also it would kinda suck if schools banned Nspires due to people running CAS OSes on them, but again, on the 83+ there are apps and programs with 89 functions and schools still allow those calcs anyway (altough some teachers require a mem reset prior tests).

CASes are not very large at all. YACAS is ~600KB.
As for how hard it is to write one, it strongly depends on how many features you want.
*Expansion is relatively easy.
*Equation solving greatly varies depending on what equations you want to solve (easy to solve polynomials, hard to solve stuff like sqrt(x) + sqrt(x+1) = sqrt(x+2)).
*Differentiation is not too bad.
*Algebraic simplification varies depending on how powerful you want the simplifier to be.
*Symbolic matrices are easy once you have everything else done.
*Rational integration is doable (I had one in Nspire BASIC).
*Algebraic integration is very, very, very hard. The description of the algorithm fills a large   chapter in a textbook, and even then is tricky to understand.

True, altough it's them who decided to leave their old calcs open to ASM lol. They should have known in 1999 that leaving their 83+ open to ASM would allow people to write CAS softwares for them (even if not necessarly complete)

Well, the solve( command is all one would need to cheat on the ACT, but I suppose that's not really the goal.

I'm just thinking: the 92 takes about a second for the quadratic formula. How long would that translate to in the 83+s mere 6 mhz?

C works on the NSpire, but not C++. A port wouldn't be quite as easy due to changes from C to C++, but still very possible. Definitely much easier than, say, asm to Axe parser, as the languages are very similar in structure.