Tyler Marr

> @marrts Actually now that I have looked at it again. It looks like it is just storing already calculated data for reporting when the table is generated. The additional...

> @marrts Just as a precaution do you mind creating a simple google benchmark for this? Yeah, I'll do that sometime soon. Totally understand why you'd want that.

Would it make more sense to never copy the input and just create a new CI every time? And then add optional parameter(s) for min and/or max length?

Setting this to WIP while #887 is being reviewed. In my brief benchmark testing offline I think I might need to make some minor adjustments, but there are still some...

Here is a breakdown of the comparison ## Head to head comparison ``` Benchmark Name Before (ms) After (ms) Difference (ms) Percent Increase (%) --------------------------------------------------------------- ------------- ------------ ----------------- ---------------------- BM_CHECK_TRAJ_CONTINUOUS_STATE_SOLVER-TRAJ1/real_time...

> @marrts is the -Debug with the print out and the those without you expect there to be no difference because all works should only happen debug is enabled? I'm...

> Sorry, I am trying to confirm that with this feature disabled that performance is not impacted. I am assuming that the items that don't include -Debug or -D in...

Also, I think there are other opportunities for improvements. The difference between using a state solver versus a joint group is very surprising to me and is evident by every...

Also for reference, here is me running the current origin/master back-to-back, so demonstrating no code changes to get a sense of the noise in the benchmarking. ``` Benchmark Name Before...

Updated results from all your suggestions (I also added the before/after avg load from the CPU if that is helpful): ``` Benchmark Name Before (ms) After (ms) Difference (ms) Percent...