jsoup battles

Every model gets the same real bugs from the jsoup app. Hidden tests from the real fix decide the result. 2 models, 5 bugs, 1 videos with replays.

Leaderboard

5 bugs, every trial counts; fixed = bugs resolved in at least half their trials: best three models.

1. GPT-6 Luna: 3 of 5 bugs, null of null points (10 min 3 s).
2. Claude Haiku 5.5: 3 of 5 bugs, null of null points (19 min 14 s).

Videos with replays

• jsoup: 5 real bugs

Bugs

• Add `RcDataRecover` tag option
• Exclude cached children from attribute clone
• Clean up comment and incomplete tag tokenisation (#2636)
• Correct fragment ignores
• Fix attribute equals and hashcode to exclude internal (transient) data attributes

• Battles
• Benchmarks