moonbitlang / core / 6059
91%

DEFAULT BRANCH: main
Ran
Jobs 1
Files 437
Run time 1min
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst
coverage: 90.666%. Remained the same
6059

push

github

bobzhang 
test(diff): add QuickCheck property coverage

Add property tests for the diff package built on correlated pairs: each
case starts from a random small-alphabet base sequence and derives the
second sequence by random edits (insert/delete/substitute/duplicate/
move/swap), plus degenerate pairs (equal, empty, prefix/suffix) and
highly repetitive two-symbol inputs.

Properties checked:
- Myers and Patience scripts are well-formed per the documented Edit
  invariants (contiguous dual cursors covering both inputs, no
  zero-length edits, same-kind coalescing, Delete-before-Insert) and
  replaying them reproduces `new` exactly, in both directions and under
  tiny cutoffs that force the heuristic path.
- The minimal Myers insert/delete cost brackets the Levenshtein
  distance (lev <= cost <= 2*lev) and is zero iff the inputs agree.
- levenshtein_edits scripts replay and cost exactly edit_distance;
  degenerate shapes match their documented contract.
- edit_distance is a metric (symmetry, identity, bounds, triangle
  inequality on correlated triples); edit_distance_within and the
  string variants agree with the full distance.
- group() hunks apply back onto `old` as a patch reproducing `new`,
  keep every change edit, and their headers/render output match the
  documented unified-diff numbering, for context 0..3 and both
  algorithms.

The hunk property found the group(context=0) zero-length Equal bug
(#4051), fixed separately; this suite stacks on that fix.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

15998 of 17645 relevant lines covered (90.67%)

151042.95 hits per line

Jobs
ID Job ID Ran Files Coverage
1 6059.1 438
90.68
 GitHub Action Run
Source Files on build 6059