moonbitlang / core / 6216

91%

push

github

cleanup: merge duplicate match arms, fix wrong docs and messages

Simplifications (behavior-preserving):
- builtin/json: merge (Null,Null)/(True,True)/(False,False) arms in Eq
- list: merge duplicate break arms in zip and the take loop
- immut/priority_queue: merge Leaf/Branch arms in peek
- immut/vector: merge Empty/Leaf arms in pop_rightmost_leaf
- internal/regex_engine: merge the duplicate None arms in
  find_first_match
- sorted_map/sorted_set: drop local max helpers duplicating Int::max
- argparse: hoist split_long's duplicated strip_prefix into unwrap_or
- debug/printer: replace two manual Option matches with unwrap_or
- math: `if x < 0.0 { true } else { false }` -> `x < 0.0`

Doc/message fixes:
- builtin pad_start/pad_end: docs claimed character-count semantics;
  the implementation measures UTF-16 code units - document the actual
  behavior including non-BMP padding chars
- immut/sorted_set: length's doc said "Get the height of set"
- deque: remove's out-of-bounds message claimed the inclusive bound of
  insert (copy-paste); state the actual exclusive bound
- builtin/bytesview: deprecation hint missing "]" in `_[_:_]`
- builtin/uninitialized_array: abort said "start index" for end
  violations too; use the generic message its siblings use

Scope trim (2026-08-20): three debatable hunks from the original sweep
were dropped to keep only clear wins - the pretty_print info_size 19->6
arm collapse (grouped unrelated variants such as Tuple/Array with
literals by coincidental value; the flat table is easier to edit), the
regex_engine remove_matches Empty-into-self fold (needed a comment to
justify its correctness), and the translate push-loop -> comprehension
change (pure loop-style churn).

Reviewed by Codex CLI (codex-cli 0.144.1) in two rounds: the first
corrected the pad_start/pad_end doc wording for non-BMP padding chars;
re-review: "Approved; no findings. All four comments accurately match
the implementations." All other hunks approved in round... (continued)

14 of 14 new or added lines in 10 files covered. (100.0%)

16284 of 17949 relevant lines covered (90.72%)

332174.16 hits per line