user: replace O(N) loop in GetNext<T> with O(1) id lookup

[Pradyuman-aviator]

NextObject() calls mjs_nextElement() which calls GetNext() on flat model lists (actuators, sensors, flexes, meshes, etc.). The previous implementation scanned the list linearly to find child, despite every element already carrying a unique, contiguous id set at insertion time (AddObject/AddObjectDefault).

Replace the loop with a direct index using the id field, accessed via static_cast<const mjCBase*>(child)->id. This cast is safe because mjCBase_ publicly inherits from mjsElement as its first base class (same pointer address), and is already used throughout the codebase (e.g. mjs_getId in user_api.cc).

Resolves the TODO comment left in the original code.

[Pradyuman-aviator]

hi @quagla I was looking through the codebase to see where I could make a contribution and noticed your TODO in src/user/user_model.cc for GetNext<T>

I have implemented the direct id lookup you suggested to replace the O(N) loop. To do this, I safely cast the mjsElement* child to mjCBase* using the same pattern that is already used for mjs_getId in user_api.cc.

Let me know if there are any specific tests you'd like me to run or add for this! Happy to adjust if needed.

[Pradyuman-aviator]

@kevinzakka if you got time please review it Thank You!

[Pradyuman-aviator]

I'll check for those failed build check & gonna update soon

[Pradyuman-aviator]

@yuvaltassa i changed logic as validating before using the id and it gonna fall back to original loop if it seems invalid still O(1) in common cases

[Pradyuman-aviator]

@yuvaltassa all checks passed now

[Pradyuman-aviator]

Hi @quagla, thanks for the review! Just checking in
is there anything blocking the merge, or is it waiting
on something from my end?

[kevinzakka]

@Pradyuman-aviator Can you not spam please?

[Pradyuman-aviator]

@kevinzakka I sincerely apologize for the multiple pings! I am very new to opensource contribution and got a bit too eager while trying to apply my theoretical knowledge to real world codebases. I'm still learning the standard etiquette and pacing for PR reviews, so I appreciate your patience.
I will definitely hold back on the pings going forward. As a small peace offering 😅 I just opened a new, small PR that adds the missing asymmetric ellipsoid shell inertia test (resolving the taylorhowell TODO) here: #3336. Whenever the team has free time, I'd love for you to take a look!