Version: 2.x

Scope

zio.blocks.scope is a compile-time safe, zero-cost resource management library for Scala 3 (and Scala 2.13). It prevents a large class of lifetime bugs by tagging allocated values with an unnameable, scope-specific type and restricting how those values may be used.

At runtime the model stays simple:

Allocate eagerly (no lazy thunks)
Register finalizers
Run finalizers deterministically when a scope closes (LIFO order)
Collect finalizer failures into a Finalization (and throw/suppress appropriately)

Why Scope?

Most resource bugs in Scala are "escape" bugs:

storing a connection/stream in a field and using it after it was closed
capturing a resource in a closure that outlives a scope
passing a resource to code that might retain it
mixing values from different lifetimes ("which scope owns this?")

Scope addresses these with a tight design:

Feature	`zio.blocks.scope`
Compile-time leak prevention	✓ (`scope.$[A]` + `$` macro + `Unscoped` boundary)
Runtime overhead	~0 (scoped values erase to `A`)
Allocation model	Eager (allocation happens at `allocate`)
Finalization	Deterministic, LIFO, errors collected
Structured lifetime	Parent/child scopes, `lower` for explicit lifetime widening
Escape hatch	`leak` (warns)

If you've used try/finally, Using, or ZIO's Scope, this is the same problem space—but optimized for synchronous code with compile-time boundaries.

Quick start (Scala 3)

import zio.blocks.scope.*

final class Database extends AutoCloseable:
  def query(sql: String): String = s"result: $sql"
  def close(): Unit = println("db closed")

@main def quickStart(): Unit =
  val out: String =
    Scope.global.scoped { scope =>
      import scope.*

      val db: $[Database] =
        Resource.fromAutoCloseable(new Database).allocate

      // Safe access: the lambda parameter can only be used as a receiver
      $(db)(_.query("SELECT 1"))
    }

  println(out)

Key points:

allocate(...) returns a scoped value: scope.$[Database] (or $[Database] after import scope.*).
You cannot call db.query(...) directly on $[Database].
You use the $ access operator: $(db)(...) (or (scope $ db)(...) without the import).
The scoped block returns a plain String because String: Unscoped.
Finalizers run when the block exits, in LIFO order.

Core mental model

1) `Scope`: finalizers + type identity

Scope is a finalizer registry plus a unique type identity:

type $[+A] — a scope-tagged, path-dependent type (erases to A at runtime)
type Parent <: Scope / val parent: Parent — the scope hierarchy

Every scope instance defines a different $ type, so values from different scopes don't accidentally mix.

Scope.global.scoped { scope =>
  import scope.*
  val x: $[Int] = 1 // ok (in global, $[A] = A)
}

Global scope

Scope.global is the root:

In the global scope: type $[+A] = A (identity)
On the JVM: global finalizers run on shutdown via a shutdown hook
On Scala.js: there is no shutdown hook, so global finalizers are not run automatically

2) Scoped values: `scope.$[A]` / `$[A]`

A value of type scope.$[A] means:

"This is an A, but it is only valid while scope is alive."

Properties:

Zero-cost: $[A] is just A at runtime (casts/identity)
Incompatible across scopes: outer.$[A] is not inner.$[A]
Methods are hidden at the type level; you must use $ to access

Access operator: `(scope $ value)(f)`

The intended way to use a scoped value is:

(scope $ scopedValue)(a => a.method(...))

This is enforced by a macro that checks the lambda uses its parameter only in receiver position.

Allowed:

(scope $ db)(_.query("SELECT 1"))
(scope $ db)(d => d.query("a") + d.query("b"))
(scope $ db)(_.query("x").toUpperCase)
(scope $ db)(_.field) // field access is allowed

Rejected at compile time:

(scope $ db)(d => store(d))            // parameter used as an argument
(scope $ db)(d => () => d.query("x"))  // captured in a nested lambda
(scope $ db)(d => d)                   // returning the parameter
(scope $ db)(d => { val x = d; 1 })    // binding/storing the parameter itself

"Auto-unwrap" rule (`Unscoped`)

$ auto-unwraps when the result type is known to be safe data:

if B: Unscoped → (scope $ sa)(f) returns B
otherwise → it returns scope.$[B]

Scope.global.scoped { scope =>
  import scope.*

  val db: $[Database] = Resource.from[Database].allocate

  val s: String = $(db)(_.query("SELECT 1"))      // String is Unscoped => unwrapped
  val n: Int    = $(db)(_.query("x").length)      // Int is Unscoped => unwrapped
}

N-ary `$`: accessing multiple scoped values at once

When a result depends on two or more scoped values simultaneously, use the N-ary overloads (N = 2..5):

$(sa1, sa2)((v1, v2) => v1.method(v2.result()))
$(sa1, sa2, sa3)((v1, v2, v3) => v1.query(v2.key()) + v3.tag())

The same receiver-only grammar applies to every parameter: each vi may only appear as a method receiver (e.g., vi.method()). Feeding the result of one parameter to a method of another is permitted:

Scope.global.scoped { scope =>
  import scope.*
  val db:    $[Database]   = Resource.from[Database].allocate
  val cache: $[Cache]      = Resource.from[Cache].allocate

  // d1 and d2 are both receivers; d2.key() produces a plain String arg
  val result: String = $(db, cache)((d1, d2) => d1.query(d2.key()))
}

Rejected at compile time (same rules as N=1, applied to each parameter independently):

$(db, cache)((d1, d2) => d2)              // d2 returned directly
$(db, cache)((d1, d2) => store(d1))       // d1 passed as argument
$(db, cache)((d1, d2) => d1.method(d2))   // d2 as bare arg (not a receiver)
$(db, cache)((d1, d2) => () => d2.query()) // d2 captured in closure

The error messages name the offending parameter:

Parameter 2 ('d2') cannot be passed as an argument to a function or method.
Scoped values may only be used as a method receiver (e.g., d2.method()).

Infix syntax (scope $ sa) is only available for N=1. For N≥2, use unqualified syntax after import scope.*:

$(db, cache)((d, c) => d.query(c.key()))   // ✓ unqualified

For N>5, extract each value in sequence (all results are Unscoped strings/values and can be freely combined):

val q1 = $(db1)(_.query("a"))
val q2 = $(db2)(_.query("b"))
q1 + q2

3) `Resource[A]`: acquisition + finalization

A Resource[A] is a lazy description of how to acquire a value and register cleanup in a scope. Nothing happens until you call scope.allocate(resource) (or .allocate syntax).

Constructors

From the source:

Resource(value: => A) Wraps a by-name value; if it's AutoCloseable, close() is registered automatically (runtime check).
Resource.fromAutoCloseable(thunk: => A <: AutoCloseable) Type-safe helper that registers close().
Resource.acquireRelease(acquire: => A)(release: A => Unit)
Resource.shared(f: Scope => A) Memoized + reference-counted, thread-safe.
Resource.unique(f: Scope => A) Fresh instance per allocation.
Resource.from[T] and Resource.from[T](wires*) (macros) Constructor-based dependency injection (covered below).

Composition

Resource composes with:

map
flatMap
zip

Finalizers remain tied to the allocation scope; in composed resources, finalizers still run LIFO.

There are two distinct ideas:

Uniqueness: "each allocation yields a fresh instance"
- Use Resource.unique(...), or most ordinary Resource(...) / acquireRelease(...) resources.
- Each allocate runs the acquisition again and registers an independent finalizer.
Sharing: "reusing the same instance across multiple allocations"
- Use Resource.shared(...) (or wires/resources that convert to shared).
- Sharing is tied to reusing the same Resource.Shared value, not "magic caching inside a scope".
- The first allocation initializes via an OpenScope parented to Scope.global; subsequent allocations increment a reference count. When the last referencing scope closes, the shared scope is closed.

4) `Unscoped[A]`: types that may escape a scope

Unscoped[A] is a marker typeclass for pure data. It's used in two places:

Scope.scoped requires Unscoped[A] for the block's result type ⇒ prevents returning resources, closures, or scoped values.
$ auto-unwraps results of type B when B: Unscoped.

Built-in instances include primitives, String, many collections/containers, time values, java.util.UUID, and zio.blocks.chunk.Chunk (when element types are unscoped).

Deriving / defining your own instances

Scala 3 (derivation via Unscoped.derived):

import zio.blocks.scope.*

final case class Config(debug: Boolean)
object Config:
  given Unscoped[Config] = Unscoped.derived

Scala 2.13:

import zio.blocks.scope.*

final case class Config(debug: Boolean)
object Config {
  implicit val unscopedConfig: Unscoped[Config] = Unscoped.derived[Config]
}

Scope boundary example

import zio.blocks.scope.*

Scope.global.scoped { parent =>
  import parent.*

  val ok: String =
    parent.scoped { child =>
      "hello" // String is Unscoped
    }

  // Does not compile: returning a resourceful value from a scoped block
  // val leaked: Database =
  //   parent.scoped { child =>
  //     import child.*
  //     Resource.fromAutoCloseable(new Database).allocate
  //   }

  ok
}

5) `lower`: using a parent-scoped value in a child scope

Because each scope has its own $[A] type, a child cannot directly use a parent's $[A]. Use lower to retag a parent-scoped value into the child:

import zio.blocks.scope.*

Scope.global.scoped { outer =>
  import outer.*

  val db: $[Database] = Resource.fromAutoCloseable(new Database).allocate

  outer.scoped { inner =>
    import inner.*
    val innerDb: $[Database] = lower(db)
    $(innerDb)(_.query("child"))
  }
}

This is safe because parents always outlive children (child finalizers run before the parent closes).

6) `defer`: manual finalizers (+ cancellation)

Use defer to register cleanup. It returns a DeferHandle you can cancel.

import zio.blocks.scope.*

Scope.global.scoped { scope =>
  import scope.*

  val in = new java.io.ByteArrayInputStream(Array[Byte](1, 2, 3))

  val h: DeferHandle =
    defer(in.close())

  val first = in.read()
  println(first)

  // If you already cleaned up manually:
  // h.cancel() // thread-safe, idempotent
}

There is also a package-level helper that only requires a Finalizer:

import zio.blocks.scope.*

Scope.global.scoped { scope =>
  import scope.*
  given Finalizer = scope

  defer(println("cleanup")) // uses the package-level helper
}

7) `open()`: non-lexical, explicitly-managed child scopes

scoped ties lifetime to a block. open() creates a child scope you close explicitly.

The child scope is unowned (can be used from any thread)
Still linked to the parent: parent closing will also close the child
You must call close() on the handle to detach + finalize now

From Scope.global the returned type is Scope.OpenScope directly (because global $[A] = A):

import zio.blocks.scope.*

val os: Scope.OpenScope = Scope.global.open()

val db = os.scope.allocate(Resource.fromAutoCloseable(new Database))

// ... use db ...

os.close().orThrow()

Inside a child scope, open() returns $[Scope.OpenScope]. Prefer using it safely via $:

import zio.blocks.scope.*

Scope.global.scoped { parent =>
  import parent.*

  val os: $[Scope.OpenScope] = open()

  $(os) { h =>
    val child = h.scope
    val db    = child.allocate(Resource.fromAutoCloseable(new Database))

    // ...
    h.close().orThrow()
  }
}

8) Escape hatch: `leak`

Sometimes you must hand a raw value to code that cannot work with $[A]. Use leak:

import zio.blocks.scope.*

Scope.global.scoped { scope =>
  import scope.*

  val db: $[Database] = Resource.fromAutoCloseable(new Database).allocate

  val raw: Database = leak(db) // emits a compiler warning
  // thirdParty(raw)
}

leak bypasses compile-time guarantees—use only for unavoidable interop. If the type is genuinely pure data, prefer adding Unscoped so you don't need to leak.

Safety model (why leaking is prevented)

Scope's safety comes from three reinforcing layers.

1) Type barrier: scope-specific `$[A]`

Every scope has a distinct $[A] type. You cannot accidentally use values across scopes without an explicit conversion (lower for parent → child).

2) Controlled access: `$` macro restricts lambda usage

The $ operator only allows using the unwrapped value as a method/field receiver. This prevents:

returning the resource
storing it in a local val/var
passing it as an argument
capturing it in a closure

Also note: $ requires a lambda literal. Method references / variables are rejected:

// does not compile:
val f: Database => String = _.query("x")
(scope $ db)(f) // "$ requires a lambda literal ..."

3) Scope boundary rule: `scoped` requires `Unscoped[A]`

A scoped { ... } block can only return pure data (or Nothing). Resources and closures cannot escape.

Pragmatic safety. The type-level tagging prevents accidental scope misuse in normal code, but it is not a security boundary. A determined developer can bypass it via leak (which emits a compiler warning), unsafe casts (asInstanceOf), or storing scoped references in mutable state (var).

Closed-scope safety (runtime)

If a scope reference escapes its scoped { } block and an operation is attempted after closing, Scope throws IllegalStateException with a detailed, actionable error message:

allocate on a closed scope:

── Scope Error ─────────────────────────────────────────────────────────────────

  Cannot allocate resource: scope is already closed.

  Scope: Scope.Child

  What happened:
    A call to allocate was made on a scope whose finalizers have
    already run. The resource was never acquired.

  Common causes:
    • A scope reference escaped a scoped { } block (e.g. stored in a
      field, captured in a Future or passed to another thread).
    • close() was called on an OpenScope before all
      allocations inside it completed.

  Fix:
    Call allocate only inside a live scoped { } block, or before
    calling close() on an OpenScope.

    // Correct usage:
    Scope.global.scoped { scope =>
      import scope.*
      val db = allocate(Resource(new Database))
      $(db)(_.query("SELECT 1"))
    }

────────────────────────────────────────────────────────────────────────────────

open() on a closed scope gives the same treatment, explaining that no child scope was created and directing the user to call open() only on a live scope.
$ on a closed scope explains that the resource may have already been released and accessing it would be undefined behaviour.

The following operations on a closed scope do not throw:

defer — silently ignored (no-op)
scoped — runs normally but creates a born-closed child scope
lower — zero-cost cast, no closed check needed

Thread ownership rule (JVM)

Scopes created by scoped are owned by the entering thread.
Calling scoped on a scope you don't own throws IllegalStateException.
open() creates an unowned child scope (isOwner == true from any thread).

(Scala.js uses a trivial ownership model; isOwner is effectively always true.)

Usage examples (patterns)

Allocating and using a resource

import zio.blocks.scope.*

final class FileHandle(path: String) extends AutoCloseable:
  def readAll(): String = s"contents of $path"
  def close(): Unit = println(s"closed $path")

@main def fileExample(): Unit =
  Scope.global.scoped { scope =>
    import scope.*

    val h: $[FileHandle] =
      Resource(new FileHandle("data.txt")).allocate

    val contents: String =
      $(h)(_.readAll())

    println(contents)
  }

Nested scopes (child can use parent, not vice versa)

import zio.blocks.scope.*

final class Database extends AutoCloseable:
  def query(sql: String): String = s"result: $sql"
  def close(): Unit = println("db closed")

@main def nested(): Unit =
  Scope.global.scoped { parent =>
    import parent.*

    val parentDb: $[Database] = Resource.fromAutoCloseable(new Database).allocate

    val done: String =
      parent.scoped { child =>
        import child.*

        val db: $[Database] = lower(parentDb)
        println($(db)(_.query("SELECT 1")))

        val childDb: $[Database] = Resource.fromAutoCloseable(new Database).allocate
        println($(childDb)(_.query("SELECT 2")))

        // childDb cannot be returned to the parent (not Unscoped)
        "done"
      }

    println($(parentDb)(_.query("SELECT 3")))
    done
  }

Finalizers run child first, then parent.

Chaining resource acquisition (`$[Resource[A]]` + `.allocate`)

If a method returns Resource[A], $ returns a scoped Resource[A] (because Resource[A] is not Unscoped). Allocate it without leaking:

import zio.blocks.scope.*

final class Pool extends AutoCloseable:
  def lease(): Resource[Conn] = Resource.fromAutoCloseable(new Conn)
  def close(): Unit = println("pool closed")

final class Conn extends AutoCloseable:
  def query(sql: String): String = s"result: $sql"
  def close(): Unit = println("connection closed")

@main def chaining(): Unit =
  Scope.global.scoped { scope =>
    import scope.*

    val pool: $[Pool] = Resource.fromAutoCloseable(new Pool).allocate

    // $(pool)(_.lease()) : $[Resource[Conn]]
    val conn: $[Conn] =
      $(pool)(_.lease()).allocate

    val result: String =
      $(conn)(_.query("SELECT 1"))

    println(result)
  }

This .allocate comes from Scope.ScopedResourceOps (an extension on $[Resource[A]]).

Allocating a bare `Resource[A]` with `.allocate`

A plain Resource[A] also has .allocate as syntax sugar for scope.allocate(resource):

import zio.blocks.scope.*

Scope.global.scoped { scope =>
  import scope.*

  val db: $[Database] =
    Resource.fromAutoCloseable(new Database).allocate

  $(db)(_.query("SELECT 1"))
}

Classes with `Finalizer` parameters (cleanup-only capability)

If a class only needs cleanup registration, accept a Finalizer. DI macros inject it automatically.

import zio.blocks.scope.*

final case class Config(url: String)
object Config:
  given Unscoped[Config] = Unscoped.derived

final class ConnectionPool(config: Config)(using Finalizer):
  private val pool = s"pool(${config.url})"
  defer(println(s"shutdown $pool"))

val poolResource: Resource[ConnectionPool] =
  Resource.from[ConnectionPool](
    Wire(Config("jdbc://localhost"))
  )

@main def finalizerInjection(): Unit =
  Scope.global.scoped { scope =>
    import scope.*
    val pool: $[ConnectionPool] = poolResource.allocate
    ()
  }

When to prefer Finalizer over Scope:

you only need defer
you want to expose minimal power to the class

Classes with `Scope` parameters (scope injection)

If a class needs to allocate resources or create child scopes, accept a Scope:

import zio.blocks.scope.*

final case class Config(url: String)
object Config:
  given Unscoped[Config] = Unscoped.derived

final class Connection(config: Config) extends AutoCloseable:
  def query(sql: String): String = s"[${config.url}] $sql"
  def close(): Unit = println("connection closed")

final class RequestHandler(config: Config)(using scope: Scope):
  def handle(sql: String): String =
    scope.scoped { child =>
      import child.*
      val conn: $[Connection] = Resource.fromAutoCloseable(new Connection(config)).allocate
      $(conn)(_.query(sql))
    }

val handlerResource: Resource[RequestHandler] =
  Resource.from[RequestHandler](
    Wire(Config("jdbc://localhost"))
  )

@main def scopeInjection(): Unit =
  Scope.global.scoped { scope =>
    import scope.*
    val handler: $[RequestHandler] = handlerResource.allocate
    val out: String = $(handler)(_.handle("SELECT 1"))
    println(out)
  }

The Scope/Finalizer parameter can appear in any parameter list position; it's recognized specially by the derivation macros.

Dependency injection (DI) with `Wire` + `Resource.from`

Scope includes a small constructor-based DI layer built on top of zio.blocks.context.Context. For a comprehensive guide to Wire and its construction patterns, see the Wire reference page.

`Wire[-In, +Out]`: a dependency recipe

A Wire is a recipe for constructing Out from a Context[In] (and a Scope for finalization):

Wire.Shared → converts to Resource.shared (ref-counted sharing)
Wire.Unique → converts to Resource.unique (fresh instance)

Manual wire + Context

import zio.blocks.scope.*
import zio.blocks.context.Context

final case class Config(debug: Boolean)
object Config:
  given Unscoped[Config] = Unscoped.derived

val w: Wire.Shared[Boolean, Config] =
  Wire.shared[Config] // Boolean => Config

val deps: Context[Boolean] =
  Context(true)

@main def wireAndContext(): Unit =
  Scope.global.scoped { scope =>
    import scope.*

    val cfg: $[Config] =
      allocate(w.toResource(deps))

    val debug: Boolean =
      $(cfg)(_.debug)

    println(debug)
  }

import zio.blocks.scope.*

val ws = Wire.shared[Config] // shared recipe
val wu = Wire.unique[Config] // unique recipe

The difference is realized when converting to resources (toResource) and allocating.

`Resource.from[T](wires*)`: derive a whole object graph

Resource.from[T](wires*) is the primary entry point for DI. It:

uses provided wires as overrides
auto-creates missing wires for concrete classes (defaulting to shared)
rejects unmakeable/abstract types unless you provide a wire
detects cycles, duplicate providers, and subtype conflicts
generates a composed Resource[T] via flatMap chains (preserving sharing/uniqueness)

Example:

import zio.blocks.scope.*

final case class Config(url: String)
object Config:
  given Unscoped[Config] = Unscoped.derived

final class Logger:
  def info(msg: String): Unit = println(msg)

final class Database(cfg: Config) extends AutoCloseable:
  def query(sql: String): String = s"[${cfg.url}] $sql"
  def close(): Unit = println("database closed")

final class Service(db: Database, logger: Logger) extends AutoCloseable:
  def run(): Unit = logger.info(s"running with ${db.query("SELECT 1")}")
  def close(): Unit = println("service closed")

val serviceResource: Resource[Service] =
  Resource.from[Service](
    Wire(Config("jdbc:postgresql://localhost/db")) // leaf value
  )

@main def di(): Unit =
  Scope.global.scoped { scope =>
    import scope.*
    val svc: $[Service] = serviceResource.allocate
    $(svc)(_.run())
  }

Injecting traits via subtype wires

When a dependency is abstract, provide a wire for a concrete implementation:

import zio.blocks.scope.*

trait Logger:
  def info(msg: String): Unit

final class ConsoleLogger extends Logger:
  def info(msg: String): Unit = println(msg)

final class App(logger: Logger):
  def run(): Unit = logger.info("Hello!")

val appResource: Resource[App] =
  Resource.from[App](
    Wire.shared[ConsoleLogger] // satisfies Logger via subtyping
  )

@main def traitInjection(): Unit =
  Scope.global.scoped { scope =>
    import scope.*
    val app: $[App] = appResource.allocate
    $(app)(_.run())
  }

import zio.blocks.scope.*

trait Service
final class LiveService extends Service
final class NeedsService(s: Service)
final class NeedsLive(l: LiveService)
final class App(a: NeedsService, b: NeedsLive)

val appResource: Resource[App] =
  Resource.from[App](
    Wire.shared[LiveService]
  )
// LiveService instantiations: 1

Common runtime errors (and what they mean)

These IllegalStateExceptions are thrown when a scope operation is attempted on a closed scope. Each message identifies the scope type, explains what went wrong, lists common causes, and shows a correct usage example.

`allocate` on a closed scope

── Scope Error ─────────────────────────────────────────────────────────────────

  Cannot allocate resource: scope is already closed.

  Scope: Scope.Child

  What happened:
    A call to allocate was made on a scope whose finalizers have
    already run. The resource was never acquired.

  Common causes:
    • A scope reference escaped a scoped { } block (e.g. stored in a
      field, captured in a Future or passed to another thread).
    • close() was called on an OpenScope before all
      allocations inside it completed.

  Fix:
    Call allocate only inside a live scoped { } block, or before
    calling close() on an OpenScope.

    // Correct usage:
    Scope.global.scoped { scope =>
      import scope.*
      val db = allocate(Resource(new Database))
      $(db)(_.query("SELECT 1"))
    }

────────────────────────────────────────────────────────────────────────────────

`open()` on a closed scope

── Scope Error ─────────────────────────────────────────────────────────────────

  Cannot open child scope: scope is already closed.

  Scope: Scope.Child

  What happened:
    A call to open() was made on a scope whose finalizers have
    already run. No child scope was created.

  Common causes:
    • A scope reference escaped a scoped { } block and open()
      was called after the block exited.
    • close() was called on the parent OpenScope before
      open() was called on it.

  Fix:
    Call open() only on a live (not yet closed) scope.

    // Correct usage:
    Scope.global.scoped { scope =>
      import scope.*
      val child = open()
      $(child)(_.scope.allocate(Resource(new Database)))
    }

────────────────────────────────────────────────────────────────────────────────

`$` on a closed scope

── Scope Error ─────────────────────────────────────────────────────────────────

  Cannot access scoped value: scope is already closed.

  Scope: Scope.Child

  What happened:
    The $ operator was called on a scope whose finalizers have
    already run. The underlying resource may have been released.
    Accessing it would be undefined behavior.

  Common causes:
    • A $[A] value or its owning scope escaped a scoped { }
      block (e.g. captured in a Future, stored in a field, or
      passed to another thread).
    • close() was called on an OpenScope that still has
      live $[A] values being accessed.

  Fix:
    Ensure all $ calls occur strictly within the scoped { }
    block that owns the value, and that the scope has not been closed.

    // Correct usage:
    Scope.global.scoped { scope =>
      import scope.*
      val db = allocate(Resource(new Database))
      $(db)(_.query("SELECT 1"))  // $ used inside the block
    }

────────────────────────────────────────────────────────────────────────────────

Common compile errors (and what they mean)

This module produces two kinds of compile-time feedback:

Plain macro aborts for unsafe $ usage
ASCII-rendered errors/warnings for DI derivation + leak warnings (via internal.ErrorMessages)

Unsafe use inside `$`

All messages name the offending parameter by its 1-based index and source name, and end with the receiver-only reminder. Typical messages:

Parameter 1 ('d') cannot be passed as an argument to a function or method.
Scoped values may only be used as a method receiver (e.g., d.method()).

Parameter 1 ('d') must only be used as a method receiver.
It cannot be returned, stored, passed as an argument, or captured.
Scoped values may only be used as a method receiver (e.g., d.method()).

Parameter 1 ('d') cannot be captured in a nested lambda, def, or anonymous class.
Scoped values may only be used as a method receiver (e.g., d.method()).

Parameter 2 ('cache') cannot be passed as an argument to a function or method.
Scoped values may only be used as a method receiver (e.g., cache.method()).

$ requires a lambda literal, e.g. $(x)(a => a.method()).
Method references and variables are not supported.

Not a class (`Wire.shared/unique` on a trait / abstract)

── Scope Error ─────────────────────────────────────────────────────────────────

  Cannot derive Wire for MyTrait: not a class.

  Hint: Use Wire.Shared / Wire.Unique directly.

───────────────────────────────────────────────────────────────────────────────

No primary constructor

── Scope Error ─────────────────────────────────────────────────────────────────

  MyType has no primary constructor.

  Hint: Use Wire.Shared / Wire.Unique directly
        with a custom construction strategy.

───────────────────────────────────────────────────────────────────────────────

`Resource.from[T]` used when `T` has dependencies

── Scope Error ─────────────────────────────────────────────────────────────────

  Resource.from[MyService] cannot be derived.

  MyService has dependencies that must be provided:
    • Config
    • Logger

  Hint: Use Resource.from[MyService](wire1, wire2, ...)
        to provide wires for all dependencies.

───────────────────────────────────────────────────────────────────────────────

Unmakeable type (primitives, functions, collections)

── Scope Error ─────────────────────────────────────────────────────────────────

  Cannot auto-create String

  This type (primitive, collection, or function) cannot be auto-created.

  Required by:
  ├── Config
    └── App

  Fix: Provide Wire(value) with the desired value:

    Resource.from[...](
      Wire(...),  // provide a value for String
      ...
    )

───────────────────────────────────────────────────────────────────────────────

Abstract type (trait / abstract class dependency)

── Scope Error ─────────────────────────────────────────────────────────────────

  Cannot auto-create Logger

  This type is abstract (trait or abstract class).

  Required by:
  └── App

  Fix: Provide a wire for a concrete implementation:

    Resource.from[...](
      Wire.shared[ConcreteImpl],  // provides Logger
      ...
    )

───────────────────────────────────────────────────────────────────────────────

Duplicate providers (ambiguous wires)

── Scope Error ────────────────────────────────────────────────────────────────

  Multiple providers for Service

  Conflicting wires:
    1. LiveService
    2. TestService

  Hint: Remove duplicate wires or use distinct wrapper types.

───────────────────────────────────────────────────────────────────────────────

Dependency cycle

── Scope Error ────────────────────────────────────────────────────────────────

  Dependency cycle detected

  Cycle:
    ┌───────────┐
    │           ▼
    A ──► B ──► C
    ▲           │
    └───────────┘

  Break the cycle by:
    • Introducing an interface/trait
    • Using lazy initialization
    • Restructuring dependencies

───────────────────────────────────────────────────────────────────────────────

── Scope Error ────────────────────────────────────────────────────────────────

  Dependency type conflict in MyService

  FileInputStream is a subtype of InputStream.

  When both types are dependencies, Context cannot reliably distinguish
  them. The more specific type may be retrieved when the more general
  type is requested.

  To fix this, wrap one or both types in a distinct wrapper:

    case class WrappedInputStream(value: InputStream)
  or
    opaque type WrappedInputStream = InputStream

───────────────────────────────────────────────────────────────────────────────

Duplicate parameter types in a constructor

── Scope Error ────────────────────────────────────────────────────────────────

  Constructor of App has multiple parameters of type String

  Context is type-indexed and cannot supply distinct values for the same type.

  Fix: Wrap one parameter in an opaque type to distinguish them:

    opaque type FirstString = String
  or
    case class FirstString(value: String)

───────────────────────────────────────────────────────────────────────────────

Leak warning

── Scope Warning ───────────────────────────────────────────────────────────────

  leak(db)
       ^
       |

  Warning: db is being leaked from scope zio.blocks.scope.Scope.Child[...].
  This may result in undefined behavior.

  Hint:
     If you know this data type is not resourceful, then add an Unscoped
     instance for it so you do not need to leak it.

───────────────────────────────────────────────────────────────────────────────

API reference (from source)

Examples below use Scala 3 syntax. Scala 2.13 has equivalent APIs, but macro signatures differ slightly (notably $'s return type encoding).

`Scope`

sealed abstract class Scope extends Finalizer with ScopeVersionSpecific

Associated types and hierarchy:

type $[+A]
type Parent <: Scope
val parent: Parent
def isClosed: Boolean
def isOwner: Boolean

Core operations:

def scoped[A](f: (child: Scope.Child[this.type]) => A)(using Unscoped[A]): A

def allocate[A](resource: Resource[A]): $[A]
def allocate[A <: AutoCloseable](value: => A): $[A]

// N=1 (infix available: `scope $ sa`)
infix transparent inline def $[A, B](sa: $[A])(inline f: A => B): B | $[B]

// N=2..5 (unqualified syntax: `$(sa1, sa2)(f)` after `import scope.*`)
transparent inline def $[A1, A2, B](sa1: $[A1], sa2: $[A2])(inline f: (A1, A2) => B): B | $[B]
transparent inline def $[A1, A2, A3, B](sa1: $[A1], sa2: $[A2], sa3: $[A3])(inline f: (A1, A2, A3) => B): B | $[B]
transparent inline def $[A1, A2, A3, A4, B](sa1: $[A1], sa2: $[A2], sa3: $[A3], sa4: $[A4])(inline f: (A1, A2, A3, A4) => B): B | $[B]
transparent inline def $[A1, A2, A3, A4, A5, B](sa1: $[A1], sa2: $[A2], sa3: $[A3], sa4: $[A4], sa5: $[A5])(inline f: (A1, A2, A3, A4, A5) => B): B | $[B]

def lower[A](value: parent.$[A]): $[A]

override def defer(f: => Unit): DeferHandle

def open(): $[Scope.OpenScope]

inline def leak[A](inline sa: $[A]): A

Notes:

$ (all arities) requires a lambda literal and enforces safe receiver-only usage at compile time.
$ returns B if Unscoped[B] exists; otherwise returns $[B].
N=1 is infix; N≥2 are not — use unqualified syntax after import scope.*.
For N>5, call $ once per resource and combine the resulting plain (Unscoped) values.
If the scope is closed, $, allocate, and open throw IllegalStateException with a detailed error message. defer and lower are unaffected.

Syntax enrichments available after import scope.* inside a scope:

implicit class ScopedResourceOps[A](sr: $[Resource[A]]):
  def allocate: $[A]

implicit class ResourceOps[A](r: Resource[A]):
  def allocate: $[A]

`Scope.global`

object Scope:
  object global extends Scope

Properties:

type $[+A] = A (identity)
isOwner always returns true
JVM: finalizers run at shutdown via a shutdown hook
Scala.js: shutdown hook is not available

`Scope.OpenScope`

case class OpenScope(scope: Scope, close: () => Finalization)

scope: the child scope
close(): detaches from parent, runs child finalizers (LIFO), returns Finalization

`Finalizer`

trait Finalizer:
  def defer(f: => Unit): DeferHandle

A minimal capability interface for registering cleanup.

Also available as a package-level helper:

def defer(finalizer: => Unit)(using fin: Finalizer): DeferHandle

`DeferHandle`

abstract class DeferHandle:
  def cancel(): Unit

cancel() is thread-safe and idempotent
cancellation is O(1) (true removal from a concurrent map)

`Finalization`

final class Finalization(val errors: zio.blocks.chunk.Chunk[Throwable]):
  def isEmpty: Boolean
  def nonEmpty: Boolean
  def orThrow(): Unit
  def suppress(initial: Throwable): Throwable

object Finalization:
  val empty: Finalization
  def apply(errors: Chunk[Throwable]): Finalization

`Resource[+A]`

sealed trait Resource[+A]:
  def map[B](f: A => B): Resource[B]
  def flatMap[B](f: A => Resource[B]): Resource[B]
  def zip[B](that: Resource[B]): Resource[(A, B)]

Companion constructors:

object Resource:
  def apply[A](value: => A): Resource[A]
  def fromAutoCloseable[A <: AutoCloseable](thunk: => A): Resource[A]
  def acquireRelease[A](acquire: => A)(release: A => Unit): Resource[A]
  def shared[A](f: Scope => A): Resource[A]
  def unique[A](f: Scope => A): Resource[A]

  inline def from[T]: Resource[T]
  inline def from[T](inline wires: Wire[?, ?]*): Resource[T]

Notes:

Resource.from[T] (no args) only works when T has no non-scope dependencies (constructor params may include Scope/Finalizer).
Use Resource.from[T](wires*) to provide/override dependencies and derive the full graph.

`Wire[-In, +Out]`

sealed trait Wire[-In, +Out]:
  def isShared: Boolean
  def isUnique: Boolean = !isShared

  def shared: Wire.Shared[In, Out]
  def unique: Wire.Unique[In, Out]

  def toResource(deps: zio.blocks.context.Context[In]): Resource[Out]

Wires:

object Wire:
  final case class Shared[-In, +Out](makeFn: (Scope, Context[In]) => Out) extends Wire[In, Out]
  final case class Unique[-In, +Out](makeFn: (Scope, Context[In]) => Out) extends Wire[In, Out]

  def apply[T](t: T): Wire.Shared[Any, T]

  transparent inline def shared[T]: Wire.Shared[?, T]
  transparent inline def unique[T]: Wire.Unique[?, T]

Notes:

Wire(t) wraps a pre-existing value; if it's AutoCloseable, close() is registered automatically when used.

`Unscoped[A]`

trait Unscoped[A]

object Unscoped:
  inline given derived[A](using scala.deriving.Mirror.Of[A]): Unscoped[A]
  // plus many built-in givens (primitives, collections, time, UUID, Chunk, ...)

Practical guidance (summary)

Allocate in a scope: resource.allocate (inside Scope.global.scoped { scope => import scope.* ... })
Access one scoped value: $(sa)(v => v.method()) — parameter can only be a receiver
Access two or more scoped values simultaneously: $(sa1, sa2)((v1, v2) => v1.method(v2.result())) (N=2..5)
For N>5: call $ once per resource, combine the plain results
Return only Unscoped data from scoped blocks
Use lower to use parent values inside a child
If $ returns $[Resource[A]], call .allocate on it (scoped resource chaining)
Use open() for explicitly-managed, cross-thread capable scopes
Use leak only when interop forces it; prefer Unscoped for pure data

Why Scope?
Quick start (Scala 3)
Core mental model
Safety model (why leaking is prevented)
Usage examples (patterns)
Dependency injection (DI) with Wire + Resource.from
- Wire[-In, +Out]: a dependency recipe
- Resource.from[T](wires*): derive a whole object graph
Common runtime errors (and what they mean)
Common compile errors (and what they mean)
API reference (from source)
Practical guidance (summary)

Why Scope?​

Quick start (Scala 3)​

Core mental model​

1) Scope: finalizers + type identity​

Global scope​

2) Scoped values: scope.$[A] / $[A]​

Access operator: (scope $ value)(f)​

"Auto-unwrap" rule (Unscoped)​

N-ary $: accessing multiple scoped values at once​

3) Resource[A]: acquisition + finalization​

Constructors​

Composition​

Sharing vs uniqueness (important)​

4) Unscoped[A]: types that may escape a scope​

Deriving / defining your own instances​

Scope boundary example​

5) lower: using a parent-scoped value in a child scope​

6) defer: manual finalizers (+ cancellation)​

7) open(): non-lexical, explicitly-managed child scopes​

8) Escape hatch: leak​

Safety model (why leaking is prevented)​

1) Type barrier: scope-specific $[A]​

2) Controlled access: $ macro restricts lambda usage​

3) Scope boundary rule: scoped requires Unscoped[A]​

Closed-scope safety (runtime)​

Thread ownership rule (JVM)​

Usage examples (patterns)​

Allocating and using a resource​

Nested scopes (child can use parent, not vice versa)​

Chaining resource acquisition ($[Resource[A]] + .allocate)​

Allocating a bare Resource[A] with .allocate​

Classes with Finalizer parameters (cleanup-only capability)​

Classes with Scope parameters (scope injection)​

Dependency injection (DI) with Wire + Resource.from​

Wire[-In, +Out]: a dependency recipe​

Manual wire + Context​

Sharing vs uniqueness at the wire level​

Resource.from[T](wires*): derive a whole object graph​

Injecting traits via subtype wires​

Diamond patterns share a single instance (when appropriate)​

Common runtime errors (and what they mean)​

allocate on a closed scope​

open() on a closed scope​

$ on a closed scope​

Common compile errors (and what they mean)​

Unsafe use inside $​

Not a class (Wire.shared/unique on a trait / abstract)​

No primary constructor​

Resource.from[T] used when T has dependencies​

Unmakeable type (primitives, functions, collections)​

Abstract type (trait / abstract class dependency)​

Duplicate providers (ambiguous wires)​

Dependency cycle​

Subtype conflict (related dependency types)​

Duplicate parameter types in a constructor​

Leak warning​

API reference (from source)​

Scope​

Scope.global​

Scope.OpenScope​

Finalizer​

DeferHandle​

Finalization​

Resource[+A]​

Wire[-In, +Out]​

Unscoped[A]​

Practical guidance (summary)​

ZIO Chat

Why Scope?

Quick start (Scala 3)

Core mental model

1) `Scope`: finalizers + type identity

Global scope

2) Scoped values: `scope.$[A]` / `$[A]`

Access operator: `(scope $ value)(f)`

"Auto-unwrap" rule (`Unscoped`)

N-ary `$`: accessing multiple scoped values at once

3) `Resource[A]`: acquisition + finalization

Constructors

Composition

Sharing vs uniqueness (important)

4) `Unscoped[A]`: types that may escape a scope

Deriving / defining your own instances

Scope boundary example

5) `lower`: using a parent-scoped value in a child scope

6) `defer`: manual finalizers (+ cancellation)

7) `open()`: non-lexical, explicitly-managed child scopes

8) Escape hatch: `leak`

Safety model (why leaking is prevented)

1) Type barrier: scope-specific `$[A]`

2) Controlled access: `$` macro restricts lambda usage

3) Scope boundary rule: `scoped` requires `Unscoped[A]`

Closed-scope safety (runtime)

Thread ownership rule (JVM)

Usage examples (patterns)

Allocating and using a resource

Nested scopes (child can use parent, not vice versa)

Chaining resource acquisition (`$[Resource[A]]` + `.allocate`)

Allocating a bare `Resource[A]` with `.allocate`

Classes with `Finalizer` parameters (cleanup-only capability)

Classes with `Scope` parameters (scope injection)

Dependency injection (DI) with `Wire` + `Resource.from`

`Wire[-In, +Out]`: a dependency recipe

Manual wire + Context

Sharing vs uniqueness at the wire level

`Resource.from[T](wires*)`: derive a whole object graph

Injecting traits via subtype wires

Diamond patterns share a single instance (when appropriate)

Common runtime errors (and what they mean)

`allocate` on a closed scope

`open()` on a closed scope

`$` on a closed scope

Common compile errors (and what they mean)

Unsafe use inside `$`

Not a class (`Wire.shared/unique` on a trait / abstract)

No primary constructor

`Resource.from[T]` used when `T` has dependencies

Unmakeable type (primitives, functions, collections)

Abstract type (trait / abstract class dependency)

Duplicate providers (ambiguous wires)

Dependency cycle

Subtype conflict (related dependency types)

Duplicate parameter types in a constructor

Leak warning

API reference (from source)

`Scope`

`Scope.global`

`Scope.OpenScope`

`Finalizer`

`DeferHandle`

`Finalization`

`Resource[+A]`

`Wire[-In, +Out]`

`Unscoped[A]`

Practical guidance (summary)