Repository navigation
refactor(ets): follow-ups to fixjoint (#740) #825
Description
Follow-up to #740, which added ETS.fixjoint and ETS2.fixjoint (closes #724). The feature works and is tested; these are the tidy-ups found in review, none of them blocking.
Help wanted, all small:
-
fixjointis the same code inETS.pyandETS2.py. Move it once intoBaseETS(ets/_ETS.py), with the tests shared, so a fix is made in one place. - Add a worked example to the docstring, under
.. runblock:: pycon, showing a flipped joint,unit="deg", and a robot'sets()with a joint fixed. Add.. versionadded::too. Done in docs(ets): add fixjoint examples and version markers #877. - Add
fixjointto the Modification tables on the ETS documentation pages (docs/source/ets_3d.rstandets_2d.rst) and to the lists of members in the Details sections. - Reject a
boolas the joint index.isinstance(True, int)is true, sofixjoint(True, 0.1)currently fixes joint 1. Done in fix(ets): reject boolean joint indices in fixjoint #836. - Check what happens to a joint with explicit
jindexvalues that leave a gap after fixing (the surviving indices are kept, as forsplit), and say so in the docstring; this ties in with ETS/ETS2: one joint-index resolver; the all-or-none, unique guarantee breaks under concatenation and insert #803, wherejindicesthat are not a permutation of0..n-1are not handled consistently. - Use it: an IK example that solves a 7-joint arm with one joint locked, in the IK documentation.
Activity
daoshi1593 commentedon Oct 18, 2026
I'd like to take the boolean joint-index validation item in #825: reject True and False in ETS.fixjoint and ETS2.fixjoint, with focused CPU regression tests and unchanged integer-index behavior. I checked the current issue discussion and open PRs and did not find overlapping work on this item. I'll keep the other follow-ups out of scope.
petercorke commentedon Oct 19, 2026
huxiaoyi-ovo commentedon Oct 21, 2026
Hi Peter, I'd like to take the remaining
fixjointdocumentation follow-up in #825.I'll add runnable examples for
ETS.fixjoint()andETS2.fixjoint(), covering flipped joints, degree-based values, and locking a joint in a robot's ETS, along with theversionaddeddirectives.I've already prepared the changes and verified that the full cross-platform test matrix passes. I'll keep this strictly documentation-only, separate from the boolean-index fix in #836.
Thanks!