Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[core] introduce Async.memoize #969

Merged
merged 1 commit into from
Jan 7, 2025
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
48 changes: 48 additions & 0 deletions kyo-core/shared/src/main/scala/kyo/Async.scala
Original file line number Diff line number Diff line change
Expand Up @@ -696,6 +696,54 @@ object Async:
(s(0).asInstanceOf[A1], s(1).asInstanceOf[A2], s(2).asInstanceOf[A3], s(3).asInstanceOf[A4])
}

/** Creates a memoized version of a computation.
*
* Returns a function that will cache the result of the first execution and return the cached result for subsequent calls. During
* initialization, only one fiber will execute the computation while others wait for the result. If the first execution fails, the
* cache is cleared and the computation will be retried on the next invocation. Note that successful results are cached indefinitely,
* so use this for stable values that won't need refreshing.
*
* Unlike `Memo`, this memoization is optimized for performance and can be safely used in hot paths. If you're memoizing global
* initialization code or need more control over cache isolation, consider using `Memo` instead.
*
* WARNING: If the initial computation never completes (e.g., hangs indefinitely), all subsequent calls will be permanently blocked
* waiting for the result. Ensure the computation can complete in a reasonable time or introduce a timeout via `Async.timeout`.
*
* @param v
* The computation to memoize
* @return
* A function that returns the memoized computation result
*/
def memoize[A: Flat, S](v: A < S)(using Frame): (() => A < (S & Async)) < Async =
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Can this become A < (S & Async) < Async?

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It's possible but the user would need to call flatten to use it. The anon function seems more ergonomic

IO.Unsafe {
val ref = AtomicRef.Unsafe.init(Maybe.empty[Promise.Unsafe[Nothing, A]])
() =>
@tailrec def loop(): A < (S & Async) =
ref.get() match
case Present(v) => v.safe.get
case Absent =>
val promise = Promise.Unsafe.init[Nothing, A]()
if ref.compareAndSet(Absent, Present(promise)) then
Abort.run(v).map { r =>
IO.Unsafe {
if !r.isSuccess then
ref.set(Absent)
promise.completeDiscard(r)
Abort.get(r)
}
}.pipe(IO.ensure {
IO.Unsafe {
if !promise.done() then
ref.set(Absent)
}
})
else
loop()
end if
loop()

}

/** Converts a Future to an asynchronous computation.
*
* This method allows integration of existing Future-based code with Kyo's asynchronous system. It handles successful completion and
Expand Down
107 changes: 107 additions & 0 deletions kyo-core/shared/src/test/scala/kyo/AsyncTest.scala
Original file line number Diff line number Diff line change
Expand Up @@ -1069,4 +1069,111 @@ class AsyncTest extends Test:
}
}

"memoize" - {
"caches successful results" in run {
for
counter <- AtomicInt.init(0)
memoized <- Async.memoize {
counter.incrementAndGet.map(_ => 42)
}
v1 <- memoized()
v2 <- memoized()
v3 <- memoized()
count <- counter.get
yield
assert(v1 == 42)
assert(v2 == 42)
assert(v3 == 42)
assert(count == 1)
}

"retries after failure" in run {
for
counter <- AtomicInt.init(0)
memoized <- Async.memoize {
counter.incrementAndGet.map { count =>
if count == 1 then throw new RuntimeException("First attempt fails")
else 42
}
}
r1 <- Abort.run(memoized())
v2 <- memoized()
v3 <- memoized()
count <- counter.get
yield
assert(r1.isPanic)
assert(v2 == 42)
assert(v3 == 42)
assert(count == 2)
}

"works with async operations" in run {
for
counter <- AtomicInt.init(0)
memoized <- Async.memoize {
for
_ <- Async.sleep(1.millis)
count <- counter.incrementAndGet
yield count
}
v1 <- memoized()
v2 <- memoized()
v3 <- memoized()
count <- counter.get
yield
assert(v1 == 1)
assert(v2 == 1)
assert(v3 == 1)
assert(count == 1)
}

"handles concurrent access" in run {
for
counter <- AtomicInt.init(0)
memoized <- Async.memoize {
for
_ <- Async.sleep(1.millis)
count <- counter.incrementAndGet
yield count
}
results <- Async.parallel(
memoized(),
memoized(),
memoized()
)
count <- counter.get
yield
assert(results._1 == 1)
assert(results._2 == 1)
assert(results._3 == 1)
assert(count == 1)
}

"handles interruption during initialization" in run {
for
counter <- AtomicInt.init(0)
started <- Latch.init(1)
sleeping <- Latch.init(1)
done <- Latch.init(1)
memoized <- Async.memoize {
IO.ensure(done.release) {
for
_ <- started.release
_ <- Async.sleep(50.millis)
count <- counter.incrementAndGet
yield count
}
}
fiber <- Async.run(memoized())
_ <- started.await
_ <- fiber.interrupt
_ <- done.await
v2 <- memoized()
count <- counter.get
yield
assert(v2 == 1)
assert(count == 1)
}
}

end AsyncTest
7 changes: 4 additions & 3 deletions kyo-prelude/shared/src/main/scala/kyo/Memo.scala
Original file line number Diff line number Diff line change
@@ -1,11 +1,12 @@
package kyo

/** Represents a memoization effect.
/** Represents a memoization effect for global value initialization.
*
* Memo is used to cache the results of expensive computations, allowing them to be reused without re-computation.
*
* This effect is primarily intended for initializing global values or caching results of infrequent, expensive operations. It is not
* recommended for use in hot paths or frequently executed code due to potential performance overhead.
* This effect is specifically designed for initializing global values or caching results of infrequent, expensive operations. For
* memoization in performance-sensitive code or hot paths, consider using `Async.memoize` or `kyo-cache` instead, which have lower
* overhead.
*/
opaque type Memo <: Var[Memo.Cache] = Var[Memo.Cache]

Expand Down
Loading