feat(ohno_macros): rewrite the crate from requirements and design

[Evgenii (Vaiz)]

🤖 Clawpilot here! Posted automatically by Clawpilot (an AI agent), not by a human. Please verify before acting.

What this PR does

This PR rewrites crates/ohno_macros from zero. It adds two documents that state what the crate has to deliver and how it is arranged to deliver it, then implements the crate from them.

• docs/requirements.md — the behavior the crate owes: the derive (R1), #[ohno::error] (R2), #[enrich_err] (R3), diagnostics (R4), quality gates (R5).
• docs/design.md — the phases, the module tree, the types the phases exchange, how each generated item is produced, the diagnostics anchors, and the testing plan.
• src/ — the implementation.

The shape

The derive runs three phases: TokenStream -> parse -> Ast -> validate -> Model -> generate -> TokenStream.

• parse answers "can I read this". It decodes the crate's own attributes and reports what it cannot read.
• validate answers "is this allowed". It is the only phase that reports rule violations.
• generate returns TokenStream, not Result. It cannot fail.

That last point is the design. R4 forbids a diagnostic that points into generated code, and that is only structural if a Model which could make generation fail cannot be built. Shape splits the field list around the core instead of carrying an index, so "exactly one core" is unrepresentable. Message can only be obtained through validation, so an argument that is not rooted in a field is unrepresentable.

Where an invariant is not structural, the design says so plainly. The alignment between a conversion's initializers and the field list is a relation between two values; it is held by a private field and a single constructor, not pretended into the type system.

#[ohno::error] keeps no model and #[enrich_err] keeps no Ast, because neither has enough input to warrant one.

Diagnostics

Faults accumulate through one Errors type. Its add takes tokens rather than a Span, so R4's rule that a multi-token diagnostic uses syn::Error::new_spanned is enforced by the signature.

Accumulation runs across independent concerns. A concern whose own input failed to decode is skipped, not guessed at: a template that cannot be split reports one fault, instead of that fault plus field errors invented by repairing it. This is what the separate parse phase buys.

Tests

crates/ohno/tests/** is untouched and is the sole behavioral spec. It passes unchanged, including all twelve .stderr compile-fail snapshots.

The design predicted those snapshots would need regenerating to show accumulated errors. They did not: each fixture struct breaks exactly one rule, and separate structs are separate macro invocations. The design has been corrected to record that.

The crate also has 133 unit tests, one style per phase, none of which needs the proc-macro bridge: parse_quote! in and Ast asserted for parse, the exact diagnostic set asserted for validate, and insta snapshots of prettyplease output from a hand-built Model for generate.

cargo mutants catches every mutant in the rule-bearing modules. Two classes survive by construction and are documented: the three #[proc_macro] entry points cannot be called from a unit test at all, so only crates/ohno/tests/ kills them; and a mutant that corrupts the template scanner's index arithmetic is reported as a timeout rather than a failed assertion.

Gates

cargo check, clippy at workspace lint level with warnings as errors, cargo test on both crates, doc tests, examples, cargo doc, cargo fmt --check, cargo machete and the license-header check all pass.

Notes for review

Three places where the implementation settled a detail differently from the first draft of the design, each now recorded in it:

• Message splits literal from formatted, so a static #[display("...")] renders as a string literal and costs no allocation at run time.
• #[enrich_err] applies its message through map_err rather than through Enrichable on the result. The return type is not always a Result: an implemented Future::poll returns Poll<Result<..>>, which carries map_err but is not itself Enrichable.
• The generated Debug is deliberately not #[automatically_derived]. Dead-code analysis skips field reads inside a derived Debug, so marking it would make every field that only Debug reads look unused in the user's own crate.

One limit is stated rather than hidden: the #[enrich_err] body rewrite cannot work inside a const fn. const is re-emitted faithfully, so such a function is rejected by rustc rather than by the macro. No test exercises the combination.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 100.0%. Comparing base (c69cc2c) to head (1f80640).

Additional details and impacted files

@@           Coverage Diff            @@
##             main     #656    +/-   ##
========================================
  Coverage   100.0%   100.0%            
========================================
  Files         532      538     +6     
  Lines       59839    59676   -163     
========================================
- Hits        59839    59676   -163     

Flag	Coverage Δ
linux	91.7% <100.0%> (?)
linux-arm	91.2% <100.0%> (?)
windows	92.6% <100.0%> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ 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.

[Copilot]

Pull request overview

This PR is a full rewrite of crates/ohno_macros, introducing a phased architecture (parse → validate → generate), new internal diagnostics accumulation, and a refreshed test/snapshot setup to keep macro behavior pinned by crates/ohno’s existing UI/trybuild tests.

Changes:

• Replaces the previous ohno_macros implementation with new modules for diagnostics, marker handling, message lowering, and codegen.
• Adds/updates documentation describing requirements and design for the rewritten crate.
• Adds just recipes to run and overwrite trybuild-based diagnostic snapshots more conveniently.

Reviewed changes

Copilot reviewed 147 out of 147 changed files in this pull request and generated 7 comments.

Show a summary per file

File	Description
justfiles/basic.just	Adds trybuild and trybuild-overwrite recipes for running/refreshing compile-fail diagnostic tests.
crates/ohno_macros/src/utils.rs	Removes legacy helper macros/utilities from the previous implementation.
crates/ohno_macros/src/paths.rs	Centralizes ::ohno::... paths used by generated code.
crates/ohno_macros/src/message.rs	Introduces message lowering/rendering utilities and parsing for format-like args.
crates/ohno_macros/src/marker.rs	Implements the reserved doc-marker scheme for injected core fields.
crates/ohno_macros/src/lib.rs	Rewires proc-macro entry points to new expand(...) APIs and adds module structure.
crates/ohno_macros/src/diagnostics.rs	Adds an Errors accumulator for multi-error reporting.
crates/ohno_macros/src/derive_error/types.rs	Removes prior derive helper types from the old implementation.
crates/ohno_macros/src/enrich_err/tests.rs	Removes prior enrich_err unit tests from the old implementation.
crates/ohno_macros/src/enrich_err/test_attrs.rs	Removes prior signature/attribute preservation tests from the old implementation.
crates/ohno_macros/src/enrich_err/snapshots/ohno_macros__enrich_err__tests__the_signature_survives_untouched.snap	Adds updated snapshot for signature preservation behavior.
crates/ohno_macros/src/enrich_err/snapshots/ohno_macros__enrich_err__tests__the_body_runs_inside_a_closure.snap	Adds updated snapshot for closure-wrapped body behavior.
crates/ohno_macros/src/enrich_err/snapshots/ohno_macros__enrich_err__tests__arguments_are_passed_through_unchanged.snap	Adds updated snapshot asserting argument passthrough.
crates/ohno_macros/src/enrich_err/snapshots/ohno_macros__enrich_err__tests__an_inline_capture_goes_through_format.snap	Adds updated snapshot for inline capture formatting.
crates/ohno_macros/src/enrich_err/snapshots/ohno_macros__enrich_err__tests__an_async_function_awaits_an_async_block.snap	Adds updated snapshot for async rewrite shape.
crates/ohno_macros/src/enrich_err/snapshots/ohno_macros__enrich_err__tests__a_self_prefixed_argument_is_left_alone.snap	Adds updated snapshot ensuring self. args remain unchanged.
crates/ohno_macros/src/enrich_err/snapshots/ohno_macros__enrich_err__tests__a_non_literal_first_argument_is_rejected.snap	Adds updated snapshot for rejecting non-literal messages.
crates/ohno_macros/src/enrich_err/snapshots/ohno_macros__enrich_err__tests__a_non_function_is_rejected.snap	Adds updated snapshot for rejecting non-function inputs.
crates/ohno_macros/src/enrich_err/snapshots/ohno_macros__enrich_err__tests__a_missing_return_type_is_rejected.snap	Adds updated snapshot for rejecting missing return type.
crates/ohno_macros/src/enrich_err/snapshots/ohno_macros__enrich_err__tests__a_literal_message_renders_without_format.snap	Adds updated snapshot for literal message optimization.
crates/ohno_macros/src/enrich_err/snapshots/ohno_macros__enrich_err__tests__a_bare_attribute_names_the_function.snap	Adds updated snapshot for default message behavior.
crates/ohno_macros/src/error_attr/snapshots/ohno_macros__error_attr__tests__other_attributes_and_docs_survive.snap	Adds snapshots for #[ohno::error] rewrite behavior.
crates/ohno_macros/src/error_attr/snapshots/ohno_macros__error_attr__tests__no_constructors_is_rejected.snap	Adds snapshot for #[no_constructors] rejection under #[ohno::error].
crates/ohno_macros/src/error_attr/snapshots/ohno_macros__error_attr__tests__an_ordinary_doc_comment_is_left_alone.snap	Adds snapshot for preserving ordinary doc comments.
crates/ohno_macros/src/error_attr/snapshots/ohno_macros__error_attr__tests__a_unit_struct_becomes_a_tuple_struct.snap	Adds snapshot for unit-struct rewrite behavior.
crates/ohno_macros/src/error_attr/snapshots/ohno_macros__error_attr__tests__a_tuple_struct_gains_a_trailing_core.snap	Adds snapshot for tuple-struct rewrite behavior.
crates/ohno_macros/src/error_attr/snapshots/ohno_macros__error_attr__tests__a_non_struct_is_rejected.snap	Adds snapshot for rejecting non-struct items.
crates/ohno_macros/src/error_attr/snapshots/ohno_macros__error_attr__tests__a_named_struct_gains_a_named_core.snap	Adds snapshot for named-struct core insertion.
crates/ohno_macros/src/error_attr/snapshots/ohno_macros__error_attr__tests__a_marked_field_is_rejected.snap	Adds snapshot for rejecting #[error] on user fields under #[ohno::error].
crates/ohno_macros/src/error_attr/snapshots/ohno_macros__error_attr__tests__a_hand_written_reserved_marker_is_rejected.snap	Adds snapshot for rejecting reserved marker misuse.
crates/ohno_macros/src/error_attr/snapshots/ohno_macros__error_attr__tests__a_colliding_name_is_numbered.snap	Adds snapshot for disambiguating inserted core field names.
crates/ohno_macros/src/derive_error/ast.rs	Adds the decoded AST layer used between parse and validate.
crates/ohno_macros/src/derive_error/mod.rs	Implements the parse/validate/generate pipeline and “all-or-nothing” emission gating.
crates/ohno_macros/src/derive_error/generate/mod.rs	Adds generation phase entry point plus generator-focused tests/snapshots.
crates/ohno_macros/src/derive_error/generate/traits.rs	Implements generated trait impls (Display/Error/Enrichable/ErrorExt/Debug).
crates/ohno_macros/src/derive_error/generate/conversions.rs	Implements generated From<T> conversions and From<Infallible>.
crates/ohno_macros/src/derive_error/generate/constructors.rs	Implements generated new/caused_by constructors.
crates/ohno_macros/src/derive_error/display/template.rs	Adds template splitting/lowering logic for #[display(...)].
crates/ohno_macros/src/derive_error/display/argument.rs	Adds positional argument rooting/scoping logic for #[display(...)].
crates/ohno_macros/src/derive_error/display/mod.rs	Implements #[display(...)] lowering into Message with diagnostics.
crates/ohno_macros/src/derive_error/generate/snapshots/ohno_macros__derive_error__generate__tests__the_suppressing_flags_remove_their_items.snap	Adds/updates generator snapshot coverage.
crates/ohno_macros/src/derive_error/generate/snapshots/ohno_macros__derive_error__generate__tests__generics_thread_through_every_impl.snap	Adds/updates generator snapshot coverage.
crates/ohno_macros/src/derive_error/generate/snapshots/ohno_macros__derive_error__generate__tests__conversions_initialize_every_non_core_field.snap	Adds/updates generator snapshot coverage.
crates/ohno_macros/src/derive_error/generate/snapshots/ohno_macros__derive_error__generate__tests__a_tuple_struct_generates_positional_items.snap	Adds/updates generator snapshot coverage.
crates/ohno_macros/src/derive_error/generate/snapshots/ohno_macros__derive_error__generate__tests__a_single_field_struct_takes_no_constructor_parameters.snap	Adds/updates generator snapshot coverage.
crates/ohno_macros/src/derive_error/generate/snapshots/ohno_macros__derive_error__generate__tests__a_named_struct_generates_every_item.snap	Adds/updates generator snapshot coverage.
crates/ohno_macros/src/derive_error/generate/snapshots/ohno_macros__derive_error__generate__tests__a_message_overrides_the_default.snap	Adds/updates generator snapshot coverage.
crates/ohno_macros/src/derive_error/generate/snapshots/ohno_macros__derive_error__generate__tests__a_core_in_the_middle_keeps_declaration_order.snap	Adds/updates generator snapshot coverage.
crates/ohno_macros/src/derive_error/snapshots/ohno_macros__derive_error__tests__a_valid_input_expands_to_every_item.snap	Adds/updates derive expansion snapshot coverage.
crates/ohno_macros/src/derive_error/snapshots/ohno_macros__derive_error__tests__a_rejected_input_expands_to_diagnostics_only.snap	Adds/updates derive expansion snapshot coverage.
crates/ohno_macros/src/derive_error/snapshots/ohno_macros__derive_error__tests__a_valid_shape_with_an_invalid_template_generates_nothing.snap	Adds/updates derive expansion snapshot coverage.
crates/ohno_macros/src/derive_error/snapshots/ohno_macros__derive_error__tests__the_suppressing_flags_remove_their_items.snap	Adds/updates derive expansion snapshot coverage.
crates/ohno_macros/src/derive_error/display/snapshots/ohno_macros__derive_error__display__tests__too_few_arguments_are_reported.snap	Adds/updates display lowering snapshots.
crates/ohno_macros/src/derive_error/display/snapshots/ohno_macros__derive_error__display__tests__the_added_core_is_not_referenceable.snap	Adds/updates display lowering snapshots.
crates/ohno_macros/src/derive_error/display/snapshots/ohno_macros__derive_error__display__tests__every_fault_in_one_template_is_reported_together.snap	Adds/updates display lowering snapshots.
crates/ohno_macros/src/derive_error/display/snapshots/ohno_macros__derive_error__display__tests__escapes_are_resolved_for_a_literal_message.snap	Adds/updates display lowering snapshots.
crates/ohno_macros/src/derive_error/display/snapshots/ohno_macros__derive_error__display__tests__an_unsupported_argument_root_is_reported.snap	Adds/updates display lowering snapshots.
crates/ohno_macros/src/derive_error/display/snapshots/ohno_macros__derive_error__display__tests__an_unknown_placeholder_lists_the_available_fields.snap	Adds/updates display lowering snapshots.
crates/ohno_macros/src/derive_error/display/snapshots/ohno_macros__derive_error__display__tests__an_unknown_argument_root_is_reported.snap	Adds/updates display lowering snapshots.
crates/ohno_macros/src/derive_error/display/snapshots/ohno_macros__derive_error__display__tests__an_unconsumed_argument_is_reported.snap	Adds/updates display lowering snapshots.
crates/ohno_macros/src/derive_error/display/snapshots/ohno_macros__derive_error__display__tests__an_unbalanced_brace_stops_the_lowering.snap	Adds/updates display lowering snapshots.
crates/ohno_macros/src/derive_error/display/snapshots/ohno_macros__derive_error__display__tests__a_stray_closing_brace_stops_the_lowering.snap	Adds/updates display lowering snapshots.
crates/ohno_macros/src/derive_error/display/snapshots/ohno_macros__derive_error__display__tests__a_tuple_field_is_named_by_index.snap	Adds/updates display lowering snapshots.
crates/ohno_macros/src/derive_error/display/snapshots/ohno_macros__derive_error__display__tests__a_static_template_lowers_to_a_literal.snap	Adds/updates display lowering snapshots.
crates/ohno_macros/src/derive_error/display/snapshots/ohno_macros__derive_error__display__tests__a_self_prefixed_argument_is_reported.snap	Adds/updates display lowering snapshots.
crates/ohno_macros/src/derive_error/display/snapshots/ohno_macros__derive_error__display__tests__a_raw_identifier_is_offered_with_its_prefix.snap	Adds/updates display lowering snapshots.
crates/ohno_macros/src/derive_error/display/snapshots/ohno_macros__derive_error__display__tests__a_positional_argument_is_scoped_and_parenthesized.snap	Adds/updates display lowering snapshots.
crates/ohno_macros/src/derive_error/display/snapshots/ohno_macros__derive_error__display__tests__a_named_placeholder_becomes_a_field_access.snap	Adds/updates display lowering snapshots.
crates/ohno_macros/src/derive_error/display/snapshots/ohno_macros__derive_error__display__tests__a_format_spec_survives_lowering.snap	Adds/updates display lowering snapshots.
crates/ohno_macros/src/derive_error/display/snapshots/ohno_macros__derive_error__display__tests__a_declared_core_stays_referenceable.snap	Adds/updates display lowering snapshots.
crates/ohno_macros/src/derive_error/display/snapshots/ohno_macros__derive_error__display__tests__a_declared_core_is_offered_as_an_available_field.snap	Adds/updates display lowering snapshots.
crates/ohno_macros/README.md	Updates generated README to include #[ohno::error] in the public surface list.
crates/ohno_macros/docs/requirements.md	Adds requirements doc for crate behavior and diagnostics expectations.
crates/ohno_macros/docs/error_error.md	Updates error-field selection design notes.
crates/ohno_macros/docs/error_display.md	Updates display argument-scoping design notes.
crates/ohno_macros/Cargo.toml	Adds syn feature flag (clone-impls) needed by the new implementation.
.spelling	Adds new accepted words used in docs/comments/snapshots.

Suppressed comments (1)

justfiles/basic.just:251

• Same issue as trybuild: cargo test ... --tests -- {{ FILTER }} filters test names but still builds/runs every integration test target. When FILTER is provided, prefer --test {{ FILTER }} so only the selected trybuild target is executed while overwriting snapshots.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[Copilot]

Pull request overview

Copilot reviewed 147 out of 147 changed files in this pull request and generated no new comments.

Suppressed comments (2)

crates/ohno_macros/docs/requirements.md:9

• The requirements doc says unit tests and expansion snapshots were removed, but this rewrite adds extensive unit tests and insta snapshots under crates/ohno_macros/src/**/snapshots. This wording is currently inaccurate and could mislead readers about where internal checks live vs. what is authoritative.

The authority for behavior is `crates/ohno/tests/**` (integration tests) and
`crates/ohno/tests/ui/**` (compile-fail snapshots). Nothing in the crate's own
tree is authoritative; internal unit tests and expansion snapshots were removed
with the implementation.

crates/ohno_macros/src/error_attr/mod.rs:140

• unused_name uses .unwrap_or_default(), which can silently return an empty string if the search ever fails and then fail later (or generate a bogus ident). Since the preceding comment argues the search is guaranteed to find a free suffix, use .expect(...) so an invariant break is reported immediately and with a clear message.

[Evgenii (Vaiz)]

🤖 Clawpilot here! Posted automatically by Clawpilot (an AI agent), not by a human. Please verify before acting.

Answering the two suppressed comments in #656 (review), since they are not threads and cannot be replied to inline.

1. docs/requirements.md:9 — "unit tests and expansion snapshots were removed" is inaccurate. Agreed, and fixed in 1fd517a0. Answered in full on the thread carrying the same finding: #656 (comment)

2. src/error_attr/mod.rs:140 — unused_name should expect rather than unwrap_or_default. Declined; the branch is unreachable, and by construction rather than by convention.

unused_name returns ohno_core when that name is free. Otherwise it searches ohno_core_1 ..= ohno_core_n, where n is the number of declared fields. Reaching the search means one field is already called ohno_core, so at most n - 1 of the n candidates can collide, and find always has a free one. That is what the doc comment above it states, and it is a bound on the input rather than an assumption about it.

Swapping in expect changes an unreachable None into an unreachable panic. It would read as a real failure mode to anyone auditing the crate for panics, and it does not make the invariant any more true. The suggested benefit — a clearer message if it ever fired — is a message nobody can produce an input for.

(The third suppressed comment, on justfiles/basic.just:251, is the same claim as the inline one on line 237 and is answered there: #656 (comment))

[Copilot]

Pull request overview

Copilot reviewed 147 out of 147 changed files in this pull request and generated no new comments.

Suppressed comments (1)

crates/ohno_macros/src/error_attr/mod.rs:140

• unused_name is documented as a bounded search that always finds a free candidate, but the fallback currently uses .unwrap_or_default(). If the invariant is ever violated, this would produce an empty identifier string and then panic later (likely inside format_ident!) with a much less actionable error. Prefer an .expect(...) here to make the invariant explicit and keep any failure localized to this logic.

[martintmk]

review this PR

[martintmk]

[Copilot speaking]

Reviewed the rewrite end to end: the three-phase split is a real improvement over what it replaces, the diagnostics story is coherent, and the test design (whole-expansion snapshots, per-phase styles, no proc-macro bridge) is the right call. Two things stop it: #[enrich_err] loses a case that compiles on main today, and the public rustdoc lost consumer-facing facts the requirements doc still claims are documented.

Verdict: changes requested — one verified behavioural regression, one docs regression, plus non-blocking notes inline.

On design.md more broadly: it earns its length, the arguments are real and the "Decided" list is genuinely useful. The one place it overstates itself is the R4 guarantee — "generate cannot fail" is true of generate, but the #[display(...)] format spec is copied through unchecked, so rustc can still land an error on the derive. Inline comment on template.rs with the reproduction.

I am an AI agent; verify before acting on any of this. Every claim below was checked against a build of this branch.

[Copilot]

Pull request overview

Copilot reviewed 152 out of 152 changed files in this pull request and generated no new comments.

Suppressed comments (1)

crates/ohno_macros/src/derive_error/display/argument.rs:77

• root() treats any integer literal as a tuple-field index via base10_digits(), which strips suffixes. That means an argument like 0u8 would be accepted as index 0 during validation, but later expanded as self.0u8, which Rust does not accept for tuple-field access. Similar issue for suffixed floats used for nested tuple access (0.1f32).

Reject suffixed numeric literals up front so invalid inputs produce a macro diagnostic (anchored to the argument) instead of a syntax error in expanded code.

        Expr::Lit(literal) => match &literal.lit {
            Lit::Int(value) => Root::Index(value.base10_digits().to_owned(), expr),
            // `0.1` is a nested tuple access, and only its leading component names a field.
            Lit::Float(value) => value
                .base10_digits()

[Evgenii (Vaiz)]

🤖 Clawpilot here! Posted automatically by Clawpilot (an AI agent), not by a human. Please verify before acting.

Re the suppressed comment on derive_error/display/argument.rs:77 (suffixed numeric literals) — accepted, and fixed in cecf73af.

You are right, and it is worth spelling out why this one matters. root read every integer literal as a tuple index through base10_digits(), which strips the suffix. So #[display("{}", 0u8)] validated against field 0 — which exists on a tuple struct, so no diagnostic fired — and then scope_to_self expanded the original expression as &(self.0u8). A suffix is not part of a tuple index, so that is a syntax error pointing into generated code, which is exactly the failure mode R4 exists to prevent. The suffixed-float case (0.1f32) is the same fault by the same route.

The fix is a guard on each literal arm requiring an empty suffix, so such a root falls through to the Unsupported arm that was already there. No new diagnostic and no new message string: self.0u8 genuinely is "nothing that can legally follow self.", so the existing unsupported-root text applies verbatim.

Verified rather than assumed: the new unit test a_suffixed_numeric_literal_is_not_a_tuple_field was written first and watched to FAIL on the unfixed code (expected unsupported: 0u8) before the guard was added. Both directions of the guard are pinned — flipping it to always-true breaks the new test, always-false breaks an_integer_literal_names_a_tuple_field — and cargo mutants over the file reports no survivors and no timeouts. just clippy, just format-check, 144 ohno_macros tests, the full ohno suite including the trybuild snapshots, and 100% line coverage across all 18 files are all clean.

One doc sentence in docs/error_display.md had become imprecise as a result ("A numeric literal is the exception") and now says unsuffixed, with the 0u8 case named.

This is a good catch of the kind no gate here could make — the fault was invisible to the test suite because nothing exercised a suffixed literal, and invisible to coverage and mutation because the wrong behaviour lived in a branch that was fully covered doing the wrong thing. Thanks.

[Copilot]

Pull request overview

Copilot reviewed 152 out of 152 changed files in this pull request and generated 1 comment.

[Copilot]

Pull request overview

Copilot reviewed 152 out of 152 changed files in this pull request and generated no new comments.

Suppressed comments (1)

crates/ohno_macros/src/message.rs:54

• Message::render rebuilds string literals with Span::call_site(), and enrich_err::parse_message/display::lower convert LitStr into String early. This drops the user-authored literal span, so any compiler diagnostics originating from the generated format! (e.g., bad captures like {missing} or invalid format specs) will point at the macro expansion/call-site instead of the attribute argument that caused the error. If the design goal is for diagnostics to be anchored in user code, preserve the original LitStr span through lowering (e.g., store the span in Message and use it when constructing LitStr in render).