jackfirth / resyntax / #578

97%

master: 97%

Pull #841

cover

Stop traversing whole subtrees to decide if syntax is one line or many

`oneline-syntax?` and `multiline-syntax?` were implemented by counting every
distinct source line among all the atoms in a syntax object, using a
`syntax-search` stream fed through a rebellion transducer pipeline. Neither
caller needs an exact count, and both build a lazy stream over the entire
subtree even when the second line is found immediately.

Counting directly with a recursive traversal that stops at two distinct lines
cuts analysis time for `xrepl-lib/xrepl/xrepl.rkt` (1877 lines) from ~78 seconds
to ~65 seconds.

The traversal deliberately mirrors `syntax-search`'s shape, including its
handling of improper lists and its treatment of non-list, non-atom syntax (such
as vectors) as having no lines at all. Both implementations were run against
every subform of several large source files, plus hand-built edge cases, with
identical results on all 28,585 syntax objects.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

25 of 26 new or added lines in 1 file covered. (96.15%)

17504 of 18122 relevant lines covered (96.59%)

0.97 hits per line