Ten more hadith collections Sources now lists 18: the Quran, 16 hadith collections and Hisnul Muslim. New are MA (Muwatta), DA (ad-Darimi), RS, BM, AM, MK, SM and the three forties NW/QD/SW. Six of them have no citation numbering QQL can source, so CODE::N and the unscoped exact search are refused rather than answered from the sequential position — the idInBook trap one level out. Documented on Sources with the two forms it reaches and the three it does not. Also: ad-Darimi is Arabic only upstream, index sizes and rebuild times are remeasured (28 MB / ~45 s vectors, 22 MB / ~11 s fulltext), and the "no submodules" line now says what the two checked-in ones are for. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Data is committed directly; the submodules are gone Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Full-text indexes are committed too Both index sets now ship, so a checkout searches with no build step. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Similarity search is now *"term", not `term` Backticks are command substitution in bash, which made the old spelling hazardous to type. The marker now matches the full-text one: a prefix before an ordinary quote, either quote accepted. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Add the QQL manual Twelve pages covering the query language, sources, search, every interface, and the architecture. Every query shown has been run against the real data. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>