GeezerVison

step_info() raises IndexError when the response never reaches the rise-time limits within the time vector:

>>> import control as ct
>>> ct.step_info(ct.tf([2.99, -0.0685], [1, 2.83]))
IndexError: index 0 is out of bounds for axis 0 with size 0

This system is stable. The step response starts at 2.99, from the feedthrough, and decays to the DC gain of -0.0242. The default time vector ends at about 2.44 s, where the response is about -0.021. That is only about 87% of the final value, so np.nonzero(...)[0][0] for the 90% level is empty. The same error happens for time-series input whose yfinal is never approached, e.g. ct.step_info(1 - np.exp(-t), t, yfinal=1) with t in [0, 1].

Fix: if either rise-time level is never reached, RiseTime, SettlingMin and SettlingMax stay NaN. SettlingMin and SettlingMax are measured from the rise time onward. The other entries are computed as before. SettlingTime already falls back to NaN when the response doesn't settle. This follows the same idea as the fix for #1014.

Tests: added test_step_info_rise_not_reached to timeresp_test.py.

• It covers time-series data and the transfer function above.
• It fails without the fix with the IndexError, and passes with it.
• Full control/tests result: 3858 passed, 554 skipped (no slycot).
• ruff check is clean on the changed files.

AI-assisted (Claude): I used it to help read the code and draft the test. I reviewed and ran everything myself.