fix(data): convert self-referential container fields and defaults to declared types

[Roger-luo]

Fixes two closely-related bugs in how @data handles self-referential /
container-collapsing field types. Both stem from the fact that such fields are
stored as Any in the variant storage struct (the wrapper Type doesn't exist
yet, and self-references can't be named), so Julia's inner constructor performs
no conversion.

#32 — convert constructor arguments to declared field types

When a declared field type collapses to Any in storage (a Vector{Any} field
or a self-referential Vector{Self}), the inner constructor stored the value
unchanged while getproperty / variant_getfield type-asserted against the
declared annotation, raising a TypeError.

Both the positional and keyword constructors now Base.convert each argument to
its declared annotation type, restoring normal struct-construction semantics. The
conversion is a no-op when the value already matches, so construction stays type
stable, and it lets the explicit-brace form of self-referential constructors
promote a parametric singleton bottom (e.g. Empty()).

#33 — resolve the ADT name to .Type in default values

Inside the generated module the bare ADT name resolves to the module, so a
self-referential default like:

@data SExpr begin
    struct Add
        arguments::Vector{SExpr} = SExpr[]
    end
end

SExpr.Add()   # ERROR: MethodError: no method matching getindex(::Module)

expanded SExpr[] to getindex(::Module) and errored. The declared field type
already treated SExpr as SExpr.Type; the default value did not.

New eval_default processes a field's default like eval_global_ref but
additionally rewrites the ADT name (and its explicit SExpr.Type / SExpr{...}
forms) to the module-local Type alias, so the name means SExpr.Type in
defaults just as in field types. Qualified access such as SExpr.Add is
preserved so variant constructors stay reachable.

Tests

• test/data/emit/container.jl — regression tests for TypeError in getproperty on container types #32 (Vector{Any} and
  self-referential Vector{Self} fields across anonymous/named/keyword
  constructors; explicit-brace self-ref promotion for Singleton with parameters #34).
• test/data/emit/selfref.jl — regression tests for Clarify typing of default values in @data #33 (bare name, explicit
  .Type, and parametric SExpr{T}[] defaults).
• Full suite passes: data 310, match 121, derive 30, perf 5.

Fixes #32
Fixes #33

🤖 Generated with Claude Code

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
moshi-jl	Ready	Preview, Comment	Jul 12, 2026 7:03pm

[codecov]

Codecov Report

❌ Patch coverage is 96.00000% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 91.74%. Comparing base (a74db69) to head (d020b87).
⚠️ Report is 2 commits behind head on main.

Files with missing lines	Patch %	Lines
src/data/emit/cons.jl	96.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #85      +/-   ##
==========================================
+ Coverage   91.69%   91.74%   +0.04%     
==========================================
  Files          43       43              
  Lines        1661     1683      +22     
==========================================
+ Hits         1523     1544      +21     
- Misses        138      139       +1     

☔ 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.

[github-actions]

Benchmark Results (Julia v1)

Time benchmarks

	main	d020b87...	main / d020b87...
adt_transform/transform n=100	6.97 ± 0.15 μs	6.73 ± 0.19 μs	1.04 ± 0.037
adt_transform/transform n=1000	0.0671 ± 0.00081 ms	0.0662 ± 0.00096 ms	1.01 ± 0.019
linked_list/sum n=100	0.471 ± 0.01 μs	0.481 ± 0.001 μs	0.979 ± 0.021
linked_list/sum n=1000	5.19 ± 0.051 μs	5.13 ± 0.041 μs	1.01 ± 0.013
time_to_load	0.0772 ± 0.0019 s	0.0791 ± 0.0012 s	0.976 ± 0.028

Memory benchmarks

	main	d020b87...	main / d020b87...
adt_transform/transform n=100	0.204 k allocs: 6.34 kB	0.204 k allocs: 6.34 kB	1
adt_transform/transform n=1000	2.02 k allocs: 0.0619 MB	2.02 k allocs: 0.0619 MB	1
linked_list/sum n=100	0 allocs: 0 B	0 allocs: 0 B
linked_list/sum n=1000	0 allocs: 0 B	0 allocs: 0 B
time_to_load	0.145 k allocs: 11 kB	0.145 k allocs: 11 kB	1