[luketpeterson]

Adding a guard so a panic inside the Lattice impl leads to an abort, rather than UB. Validated no perf impact and does not interfere with catching panics that come from outside the Lattice implementation.

Addresses #185

[github-actions]

Bench A/B vs base: done

job log · 2026-11-11 12:09:55 UTC

base 8172d18 → head 7b404e7, 3 round(s), median of each run averaged; negative is faster

bench	cases	geomean	largest gain	largest loss	>5% faster	>5% slower
shakespeare	7	+0.6%	-1.3% shakespeare/shakespeare_words_get	+3.4% shakespeare/shakespeare_sentences_get	0	0
cities	5	+1.6%	-3.2% cities/cities_val_count	+13.5% cities/cities_get_act	0	1
sparse_keys	108	+0.1%	-2.6% sparse_descend_until/500	+9.1% sparse_insert/50	0	3
binary_keys	89	-3.8%	-7.8% binary_set_val_at_short_replace/2	+2.5% binary_insert/1600	14	0
superdense_keys	116	-0.1%	-11.9% superdense_drop_bench/1600	+5.6% superdense_k_path_iter/4000	4	1
act_paths	46	-0.8%	-9.4% shape_map_to_act_cata/random_long	+4.2% size_dense_paths_to_act/200000	1	0
zipper_head_owned	16	+2.9%	-4.3% zipper_head_owned/map_reader_get_focus	+26.2% zipper_head_owned/borrowed_head_read_missing_path	0	3
product_zipper	4	-1.0%	-2.9% product_zipper/introspecting_pathmap_pathmap	+0.1% product_zipper/generic_act_act	0	0

27 case(s) moved more than 5%

bench	case	base	head	change
zipper_head_owned	zipper_head_owned/borrowed_head_read_missing_path	13.66 µs	17.24 µs	+26.2%
zipper_head_owned	zipper_head_owned/borrowed_head_read_value_in_shared_parent	12.65 µs	15.14 µs	+19.7%
cities	cities/cities_get_act	674.60 ms	765.63 ms	+13.5%
superdense_keys	superdense_drop_bench/1600	4.71 µs	4.15 µs	-11.9%
act_paths	shape_map_to_act_cata/random_long	43.42 ms	39.32 ms	-9.4%
sparse_keys	sparse_insert/50	1.66 µs	1.81 µs	+9.1%
binary_keys	binary_set_val_at_short_replace/2	18 ns	17 ns	-7.8%
sparse_keys	sparse_insert/100	3.97 µs	4.26 µs	+7.4%
binary_keys	binary_zipper_prev_iter/400	354.33 µs	329.57 µs	-7.0%
binary_keys	binary_zipper_prev_iter/100	93.73 µs	87.24 µs	-6.9%
binary_keys	binary_zipper_prev_iter/200	182.70 µs	170.10 µs	-6.9%
superdense_keys	superdense_get/32000	463.77 µs	432.20 µs	-6.8%
binary_keys	binary_zipper_prev_iter/50	47.95 µs	44.74 µs	-6.7%
binary_keys	binary_val_count_bench/4000	31.99 µs	29.92 µs	-6.5%
binary_keys	binary_zipper_prev_iter/800	683.10 µs	639.40 µs	-6.4%
binary_keys	binary_zipper_prev_iter/1600	1.32 ms	1.23 ms	-6.4%
superdense_keys	superdense_drop_bench/100	53 ns	50 ns	-6.3%
binary_keys	binary_get/2000	143.00 µs	134.63 µs	-5.9%
superdense_keys	superdense_k_path_iter/4000	25.20 µs	26.61 µs	+5.6%
binary_keys	binary_meet/50	5.42 µs	5.13 µs	-5.4%
binary_keys	binary_zipper_prev_step_iter/200	310.53 µs	293.83 µs	-5.4%
sparse_keys	sparse_insert/200	8.93 µs	9.41 µs	+5.3%
superdense_keys	superdense_zipper_cursor/200	1.19 µs	1.13 µs	-5.3%
binary_keys	binary_zipper_prev_step_iter/400	597.67 µs	566.87 µs	-5.2%
binary_keys	binary_meet/1600	160.67 µs	152.47 µs	-5.1%
binary_keys	binary_drop_head/400	38.91 µs	36.92 µs	-5.1%
zipper_head_owned	zipper_head_owned/shared_parent_reader_val	1.47 µs	1.54 µs	+5.0%

Full tables per bench are in the job log and the bench-out artifact.