[gvegayon]

What

• Keeps LinkingTo: barry (>= 0.2.2) (the CRAN version) so CI can resolve dependencies. The motif formula parser fix (Fix defm_motif_parser reading covariate names as motif terms USCbiostats/barry#26, synced in Sync headers from barry: fix defm formula parser covariate handling USCbiostats/barryr#12) applies when defm is built against barry >= 0.2.2.9000.
• Adds inst/tinytest/test_formula_parser.R with regression tests.
• Documents the covariate syntax in td_formula() (roxygen + regenerated man/defm_terms.Rd).
• Bumps the version to 0.2.2.9000 and adds a NEWS entry.

Why

td_formula() could build a different model from the one written, with no error:

Formula (order 1, Y = a, b)	Before	After
{y0} > {y0} x Day1	Motif {a+}>{a+, b+} x Day1	Motif {a+}>{a+} x Day1
{y0} > {y1}x Female	Motif {a+}>{b+} (covariate dropped)	Motif {a+}>{b+} x Female
{y0_1} > {y1}	Motif {a+, b+} (not a transition)	error
{y0} x Day5	error: column out of range	works

The root cause: barry looked for y<col> tokens across the whole formula, including the covariate name.

Reviewer notes

• Merge order: this PR can merge on its own. The new regression tests (test_formula_parser.R) first check which parser defm was compiled with, by trying {y0} > {y1} x Day1 (the old parser throws on it). If it fails they call exit_file(). This avoids depending on the installed barry version, which may differ from the headers defm was built against. Once Fix defm_motif_parser reading covariate names as motif terms USCbiostats/barry#26 and Sync headers from barry: fix defm formula parser covariate handling USCbiostats/barryr#12 are merged, CI installs barryr from GitHub (Remotes) and the tests run. I checked both paths locally: CRAN barry 0.2.2 gives 25/25 with the parser tests skipped, and barry 0.2.2.9000 gives 30/30.
• Consider raising LinkingTo to barry (>= 0.2.3) once that is on CRAN.
• Behavior change: "{...} xCovar" (no whitespace after x) is now a syntax error instead of being ambiguous.
• Existing fitted models that used a covariate with y<digit> in its name, or }x, were estimating a different motif than intended and should be re-run.
• Tests: tinytest::test_package("defm") gives 30/30 passing against the barryr branch.

🤖 Generated with Claude Code

[github-actions]

Thank you for your contribution @gvegayon 🚀! The pkgdown site preview is ready for review 👉 Download here 👈!
_{(The artifact expires on 2026-12-25T01:42:58Z. You can re-generate it by re-running the workflow here.)}

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 81.45%. Comparing base (b016e3a) to head (62e4217).
⚠️ Report is 3 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master      #12      +/-   ##
==========================================
+ Coverage   81.37%   81.45%   +0.08%     
==========================================
  Files           8        9       +1     
  Lines         451      453       +2     
==========================================
+ Hits          367      369       +2     
  Misses         84       84              

☔ 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.