fix(swift): build wire-test JSON body from typed example shape

[devin-ai-integration]

Description

Linear ticket: Closes FER-10341

Fixes 7 Swift wire-test runtime equality failures observed in the latest nightly Square baseline-e2e run — OrdersClient.{updateOrder2,payOrder2,calculateOrder2,cloneOrder2,retrieveOrder2,createOrder2} and CheckoutClient.createPaymentLink2.

Root cause

When a wire-test response body contains a Dictionary<String, T> whose value is an explicit-null Nullable<T>, the typed example shape stored on ExampleTypeReference retains the entry (mapContainer.map has a key whose value is a Container.nullable with nullable: undefined), but the IR's parallel jsonExample projection ends up with the value as JS undefined. JSON.stringify then silently drops the key from the embedded JSON body. The expected response struct, however, is built by walking the typed shape (Nullable<T>.null is preserved), so JSONDecoder.decode(...) produces [:] while the expected struct contains ["metadata": .null], and the runtime equality assertion fails:

LHS (decoded):  Nullable<Dictionary<String, Nullable<String>>>.value([:])
RHS (expected): Nullable<Dictionary<String, Nullable<String>>>.value(["metadata": Nullable.null])

Fix

Build the wire-test JSON body by walking the typed ExampleTypeReference shape directly, the same source the expected struct is built from. This keeps the two sides in sync by construction. Discriminated and undiscriminated unions (whose wire format requires type-declaration metadata not present on the example shape) fall back to the IR-supplied jsonExample, since the map-with-null-entries asymmetry does not apply within union examples.

Changes Made

• New helper buildJsonFromExampleTypeReference (in generators/swift/sdk/src/generators/test/) walks ExampleTypeReference.shape and produces a JSON-encodable JS value mirroring the wire format. Maps with explicit-null entries surface as null (not undefined) so they survive JSON.stringify.
• WireTestFunctionGenerator now routes the embedded response body through this helper instead of JSON.stringify(exampleTypeRef.jsonExample, null, 2). Existing raw-string fast path (used for the iCalendar fixture from fix(swift): emit wire-test response bodies as raw multi-line string literals #15605) is preserved.

Testing

• Unit tests added/updated — 7 new cases in buildJsonFromExampleTypeReference.test.ts covering primitives, populated and unset nullables, optionals, lists with nullable<T> items, the regression case of map<string, nullable<string>> with an explicit-null entry, and a nested-object containing such a map.
• All swift-sdk unit tests pass locally (20/20).
• CI to verify against Square baseline-e2e (the original failing fixture).

Local seed verification was not feasible because pnpm seed:build is currently broken on main due to an unrelated @fern-api/register TypeScript compile error against the pinned @fern-api/fdr-sdk (Availability enum missing Legacy/Alpha/Preview variants).

Link to Devin session: https://app.devin.ai/sessions/9531f957724c47d389f7043dd1ec80cc

[devin-ai-integration]

🤖 Devin AI Engineer

I'll be helping with this pull request! Here's what you should know:

✅ I will automatically:

• Address comments on this PR. Add '(aside)' to your comment to have me ignore it.
• Look at CI failures and help fix them

Note: I can only respond to comments from users who have write access to this repository.

⚙️ Control Options:

• Disable automatic comment and CI monitoring

[claude]

Claude Code Review

This repository is configured for manual code reviews. Comment @claude review to trigger a review and subscribe this PR to future pushes, or @claude review once for a one-time review.

_{Tip: disable this comment in your organization's Code Review settings.}

[github-actions]

SDK Generation Benchmark Results

Comparing PR branch against median of 5 nightly run(s) on main (latest: 2026-04-23T04:59:11Z).

Full benchmark table (click to expand)

Generator	Spec	main (generator)	main (E2E)	PR (generator)	Delta
swift-sdk	square	41s (n=5)	290s (n=5)	44s	+3s (+7.3%)

main (generator): generator-only time via --skip-scripts (includes Docker image build, container startup, IR parsing, and code generation — this is the same Docker-based flow customers use via fern generate). main (E2E): full customer-observable time including build/test scripts (nightly baseline, informational). Delta is computed against generator-only baseline.
⚠️ = generation exited with a non-zero exit code (timing may not reflect a successful run).
Baseline from nightly runs on main (latest: 2026-04-23T04:59:11Z). Trigger benchmark-baseline to refresh.
Last updated: 2026-05-04 22:23 UTC