Okay, just going to try to clear up some misunderstandings here.
smurfer wrote:replace bottlenecks with C/C++ code.
This post by Josh himself very clearly states that this isn't an option (actually, that's most of the point of the top half of the post). The problem was that there was
already too much C/C++ code for him to mentally handle. To continue programming, he had to offload it to other languages with simpler code.
Removing that would quite literally be rewriting the whole engine, because he'd have to strip out all the stuff that let him do that in the first place and then re-code everything that's in the files - shaders, procedural algorithms, game logic, etc.
Then, assuming he could even finish the game after going back to square one and battling the original problem (there was too much C/C++ code), he would have to take everything
back out of the game and put it exactly as it is now - undoing all the work that he did to take out the moddability in the first place. Then he'd be back where he is now: trying to figure out how to make the game both moddable, and fast.
But, as you don't care for moddability, you'd probably be pleased to know that at that point, the game would probably never be moddable. I doubt he'd try to get the moddability working again after all that. Personally I think moddability is important to help maintain a strong community and encourage growth and development. Without mods, Limit Theory would be fun only to the players that wanted LT exactly as it was released.
So, essentially:
1. Josh has problem he cannot solve; brain absolutely cannot 100k+ lines of C
2. Josh solves problem by introducing script; script simpler and easier to handle than C
3. Script permits modding. It is not the purpose, but merely a byproduct
4. Josh has problem he has not solved, but is working on solving; the game is slower with script instead of C
5. Your solution: Undo all progress from last 2.5 years, GOTO 1, try to solve the original problem which he made less headway on than this one (which is what prompted this in the first place)
To a non-programmer, it might seem trivial to roll back to an earlier version, but in actuality, so much has changed since pre-LTSL that he really would be rewriting the whole engine. That's the reality of any project. Code evolves over time, and it especially does so in a 100k+ line project. Trying to roll back to a previous version would require him to almost completely rebuild the engine, and then he would have to rebuild all the stuff in the scrips, in the engine. (And that's before putting him up against the wall that was even more impassable than this one.)
Personally I would suggest he multithread it. While that would also require him to rewrite large portions of the engine, he wouldn't run up against the 100k wall, he wouldn't have to rebuild everything in the scripts, and the game would be faster. Alternatively, he could hire another coder or two... but frankly, I don't think he has the funds left for that.
Regardless, there is no conceivable way that undoing all the work of the past 2.5 years would actually move the project forward. In fact, it would bring it to a standstill and guarantee it would never be released.