fix(server): keep union and index-signature context types across .use()

[dinwwwh]

Context added by .use() no longer breaks union or index-signature context types. Before, os.$context<{ kind: 'a', a: string } | { kind: 'b', b: number }>().use(({ next }) => next({ context: { user: 1 } })) gave the handler { kind: 'a' | 'b' }, so if (context.kind === 'a') context.a was a tsc error. With { [k: string]: unknown, db: string }, context.db became unknown. The handler now gets the full union, which narrows as expected, and db stays string.

Fixes

• Discriminated union contexts keep narrowing after .use() adds context.
• Named keys next to an index signature keep their types after .use().
• A later .use() whose middleware needs a union context keeps that union in the procedure's initial context. call(procedure, input, { context: { kind: 'a', a: 'x' } }) is no longer rejected.
• Optional and readonly markers on context keys are kept.

Performance

• No type-check cost over the old Omit. On 300 procedures with three .use() calls each: 5% fewer types and the same instantiation count (542,770 vs 543,993), with check time within noise.

Reviewer notes

• Merging into a bare object context now reads object & { user: number } instead of {} & { user: number }. The two behave the same, but three exact-type assertions changed (two in @orpc/effect).

Testing

• New type tests for unions and index signatures in context.test-d.ts and a .use() → .handler() narrowing case in builder.test-d.ts. All fail on the old code and pass with the fix.
• Root tsc over all type tests and type:check in all 28 packages pass, plus eslint.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[pkg-pr-new]

More templates

• @orpc/bun-playground
• @orpc/cloudflare-playground
• @orpc/nest-playground
• @orpc/next-playground

@orpc/ai-sdk

npm i https://pkg.pr.new/@orpc/ai-sdk@2110

@orpc/arktype

npm i https://pkg.pr.new/@orpc/arktype@2110

@orpc/bun

npm i https://pkg.pr.new/@orpc/bun@2110

@orpc/client

npm i https://pkg.pr.new/@orpc/client@2110

@orpc/cloudflare

npm i https://pkg.pr.new/@orpc/cloudflare@2110

@orpc/contract

npm i https://pkg.pr.new/@orpc/contract@2110

@orpc/experimental-effect

npm i https://pkg.pr.new/@orpc/experimental-effect@2110

@orpc/evlog

npm i https://pkg.pr.new/@orpc/evlog@2110

@orpc/hibernation

npm i https://pkg.pr.new/@orpc/hibernation@2110

@orpc/json-schema

npm i https://pkg.pr.new/@orpc/json-schema@2110

@orpc/experimental-lock

npm i https://pkg.pr.new/@orpc/experimental-lock@2110

@orpc/experimental-msw

npm i https://pkg.pr.new/@orpc/experimental-msw@2110

@orpc/nest

npm i https://pkg.pr.new/@orpc/nest@2110

@orpc/next

npm i https://pkg.pr.new/@orpc/next@2110

@orpc/node

npm i https://pkg.pr.new/@orpc/node@2110

@orpc/openapi

npm i https://pkg.pr.new/@orpc/openapi@2110

@orpc/opentelemetry

npm i https://pkg.pr.new/@orpc/opentelemetry@2110

@orpc/pinia-colada

npm i https://pkg.pr.new/@orpc/pinia-colada@2110

@orpc/pino

npm i https://pkg.pr.new/@orpc/pino@2110

@orpc/publisher

npm i https://pkg.pr.new/@orpc/publisher@2110

@orpc/ratelimit

npm i https://pkg.pr.new/@orpc/ratelimit@2110

@orpc/server

npm i https://pkg.pr.new/@orpc/server@2110

@orpc/shared

npm i https://pkg.pr.new/@orpc/shared@2110

@orpc/swr

npm i https://pkg.pr.new/@orpc/swr@2110

@orpc/tanstack-query

npm i https://pkg.pr.new/@orpc/tanstack-query@2110

@orpc/trpc

npm i https://pkg.pr.new/@orpc/trpc@2110

@orpc/valibot

npm i https://pkg.pr.new/@orpc/valibot@2110

@orpc/zod

npm i https://pkg.pr.new/@orpc/zod@2110

commit: 0553669

[codspeed]

Merging this PR will not alter performance

✅ 30 untouched benchmarks

---

_{Comparing dinwwwh:claude/union-context-narrowing-use-50db0b (0553669) with main (1d40246)}

[pullfrog]

ℹ️ Minor suggestion only — one edge case in the new distribution wrapper.

Reviewed changes

• Type-level context merge fix — replaced Omit with a homomorphic OmitKeys in both MergedContext and MergedInitialContext, so discriminated unions and named keys beside index signatures survive .use().
• Union distribution — MergedInitialContext now distributes over a union TInContext (TInContext extends any ? … : never), so each union member keeps its own extra keys instead of collapsing to the intersection of keys.
• Tests — added union/index-signature type tests in context.test-d.ts and an end-to-end .use() → .handler() narrowing test in builder.test-d.ts; updated three exact-type assertions that shifted from {} to object for bare-object contexts.

I independently confirmed pnpm type:check passes across all packages and the root type tests, that the new tests fail to compile against the old definitions (so they pin the bug), and that the index-signature override path now yields db: number rather than string & number. The fix is sound; the only note is the never fallback below.

  ^{｜ Fix all ➔ ｜ Fix 👍s ➔ ｜ View workflow run ｜ Using DeepSeek Flash (free via Pullfrog for OSS) ｜ 𝕏}

[pullfrog]

✅ No new issues found.

Reviewed changes

Since the prior pullfrog review at c0b20ccd, the only new commit (0553669b) is a refactor:

• Extracted DistributiveOmit to @orpc/shared — the private OmitKeys helper in server/context.ts is now the public DistributiveOmit<T, K extends PropertyKey> in shared/src/types.ts, with a JSDoc note and unit type tests covering plain keys, readonly/optional modifiers, union distribution, and named keys beside an index signature.
• Wired server/context.ts to the shared helper — MergedContext and MergedInitialContext now import it; the resolved types are unchanged from c0b20ccd.

The one prior finding (the never fallback in MergedInitialContext) was withdrawn: never is unreachable through .use() and was not usable on main either, so it is not a regression. pnpm type:check passes for @orpc/shared and @orpc/server.

  ^{｜ View workflow run ｜ Using DeepSeek Flash (free via Pullfrog for OSS) ｜ 𝕏}