[stevevanhooser]

Problem

did.query says depends_on(NAME, VALUE) is an exact, case-sensitive match of a dependency's name and value. Only NAME '*' is a wildcard. to_searchstructure and fieldsearch work that way.

The SQLite search did not. query_struct_to_sql_str searched meta.depends_on with LIKE "%NAME,VALUE;%", which is unanchored and ignores case. doc2sql stores that field as name1,value1;name2,value2;. As a result:

• depends_on('', V) matched any dependency name. Brute force matches none.
• depends_on('id', V) also matched subject_id,V;.
• depends_on('ID', V) matched id,V;.
• a * inside a name ('subject*') acted as a wildcard.

DID-python already matches exactly (DID-python#101), so the two languages disagreed.

Change

src/did/+did/database.m:

• New SQL test. dependsOnSqlMatch(name, value) builds the test with instr, a plain case-sensitive substring test:

  • for a named dependency: instr(';' || ifnull(doc_data.value,''), ';NAME,VALUE;') > 0
  • for '*': instr(ifnull(doc_data.value,''), ',VALUE;') > 0

  This is the same test as DID-python's _depends_on_sql_clause. It is still one test on one meta.depends_on row per document, so it costs about the same as the old LIKE.

• Negated form. ifnull makes ~depends_on include documents with no dependencies, as brute force does.

• Brute-force fallback. A NAME or VALUE that is empty, or that contains , or ;, cannot be matched exactly in the stored string. search_doc_ids sends those queries to the new brute_force_search, which runs fieldsearch on the candidate documents, or on the whole branch if there are none. Inside an AND these terms are searched last, so they only test the documents the other terms found. DID-python does the same.

src/did/+did/+datastructures/fieldsearch.m:

• hasanysubfield_* used to return early when the field held something other than a struct or cell. That skipped the negation, so ~depends_on left the document out.
• A document with no dependencies has exactly that once it is read back from a database: depends_on = [], because an empty struct array is saved as [].
• Now a non-struct value simply does not match, and ~ makes it a match. DID-python's field_search already behaved this way.

src/did/+did/query.m: the help text now states the rule.

Tests

tests/+did/+unittest/TestDependsOnParity.m uses five documents:

• subject_id→X
• id→X and element_id→Y
• ID→X
• no dependencies
• element_id→x

It runs 16 cases on the SQLite search, on fieldsearch, and inside an AND. The cases cover:

• ''
• '*'
• an exact name
• a name that is a suffix of another name (id / subject_id)
• name case and value case
• the second of two dependencies
• a name and a value that come from different dependencies
• 'subject*' is not a glob
• ~depends_on with an exact name, '*', '' and a case difference

There is also a test with two depends_on terms ANDed together. DID-python gets a matching tests/test_depends_on_parity.py with the same documents and the same expected results; DID-python main already passes it.

Compatibility

NDI-matlab only uses '*' as a wildcard, and NDI-python moved from '' to '*' (Waltham-Data-Science/NDI-python#344), so tightening '' does not affect NDI.

🤖 Generated with Claude Code

https://claude.ai/code/session_014BLjcMbQRXr5MDdivxzYEc

[github-actions]

Test Results

504 tests  +49   504 ✅ +49   1m 38s ⏱️ -1s
 41 suites + 1     0 💤 ± 0 
  1 files   ± 0     0 ❌ ± 0 

Results for commit 051f535. ± Comparison against base commit 6a398d0.

♻️ This comment has been updated with latest results.

[codecov]

Codecov Report

❌ Patch coverage is 90.69767% with 4 lines in your changes missing coverage. Please review.
✅ Project coverage is 66.91%. Comparing base (6a398d0) to head (051f535).

Files with missing lines	Patch %	Lines
src/did/+did/database.m	90.47%	4 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #235      +/-   ##
==========================================
+ Coverage   66.41%   66.91%   +0.50%     
==========================================
  Files          67       67              
  Lines        4710     4748      +38     
==========================================
+ Hits         3128     3177      +49     
+ Misses       1582     1571      -11     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.