Add OwnedStream type

Signed-off-by: Moritz Hoffmann <[email protected]>

Author: antiguru

kafkaesque/src/kafka_source.rs

@@ -1,11 +1,12 @@
 use timely::Data;
-use timely::dataflow::{Scope, Stream};
+use timely::dataflow::Scope;
 use timely::dataflow::operators::Capability;
 use timely::dataflow::operators::generic::OutputHandle;
-use timely::dataflow::channels::pushers::Tee;
 
 use rdkafka::Message;
 use rdkafka::consumer::{ConsumerContext, BaseConsumer};
+use timely::dataflow::channels::pushers::tee::PushOwned;
+use timely::dataflow::stream::OwnedStream;
 
 /// Constructs a stream of data from a Kafka consumer.
 ///
@@ -89,14 +90,14 @@ pub fn kafka_source<C, G, D, L>(
     name: &str,
     consumer: BaseConsumer<C>,
     logic: L
-) -> Stream<G, D>
+) -> OwnedStream<G, Vec<D>>
 where
     C: ConsumerContext+'static,
     G: Scope,
     D: Data,
     L: Fn(&[u8],
           &mut Capability<G::Timestamp>,
-          &mut OutputHandle<G::Timestamp, D, Tee<G::Timestamp, D>>) -> bool+'static,
+          &mut OutputHandle<G::Timestamp, D, PushOwned<G::Timestamp, Vec<D>>>) -> bool+'static,
 {
     use timely::dataflow::operators::generic::source;
     source(scope, name, move |capability, info| {
@@ -135,4 +136,4 @@ where
         }
 
     })
-}
+}

timely/examples/bfs.rs

@@ -133,7 +133,7 @@ fn main() {
                     });
                 }
             )
-            .concat(&(0..1).map(|x| (x,x)).to_stream(scope))
+            .concat((0..1).map(|x| (x,x)).to_stream(scope))
             .connect_loop(handle);
         });
     }).unwrap(); // asserts error-free execution;

timely/examples/loopdemo.rs

@@ -29,7 +29,7 @@ fn main() {
 
             let step =
             stream
-                .concat(&loop_stream)
+                .concat(loop_stream)
                 .map(|x| if x % 2 == 0 { x / 2 } else { 3 * x + 1 })
                 .filter(|x| x > &1);
             step

timely/examples/pingpong.rs

@@ -16,7 +16,7 @@ fn main() {
             (0 .. elements)
                   .filter(move |&x| (x as usize) % peers == index)
                   .to_stream(scope)
-                  .concat(&cycle)
+                  .concat(cycle)
                   .exchange(|&x| x)
                   .map_in_place(|x| *x += 1)
                   .branch_when(move |t| t < &iterations).1

timely/examples/unionfind.rs

@@ -9,6 +9,7 @@ use timely::dataflow::*;
 use timely::dataflow::operators::{Input, Exchange, Probe};
 use timely::dataflow::operators::generic::operator::Operator;
 use timely::dataflow::channels::pact::Pipeline;
+use timely::dataflow::stream::{OwnedStream, StreamLike};
 
 fn main() {
 
@@ -50,12 +51,12 @@ fn main() {
     }).unwrap(); // asserts error-free execution;
 }
 
-trait UnionFind {
-    fn union_find(self) -> Self;
+trait UnionFind<G: Scope> {
+    fn union_find(self) -> OwnedStream<G, Vec<(usize, usize)>>;
 }
 
-impl<G: Scope> UnionFind for Stream<G, (usize, usize)> {
-    fn union_find(self) -> Stream<G, (usize, usize)> {
+impl<G: Scope, S: StreamLike<G, Vec<(usize, usize)>>> UnionFind<G> for S {
+    fn union_find(self) -> OwnedStream<G, Vec<(usize, usize)>> {
 
         self.unary(Pipeline, "UnionFind", |_,_| {
 

timely/src/dataflow/channels/pushers/tee.rs

@@ -1,7 +1,7 @@
 //! A `Push` implementor with a list of `Box<Push>` to forward pushes to.
 
 use std::cell::RefCell;
-use std::fmt::{self, Debug};
+use std::fmt::{self, Debug, Formatter};
 use std::rc::Rc;
 
 use crate::dataflow::channels::{BundleCore, Message};
@@ -10,6 +10,39 @@ use crate::communication::Push;
 use crate::{Container, Data};
 
 type PushList<T, D> = Rc<RefCell<Vec<Box<dyn Push<BundleCore<T, D>>>>>>;
+/// TODO
+pub struct PushOwned<T, D>(Rc<RefCell<Option<Box<dyn Push<BundleCore<T, D>>>>>>);
+
+impl<T, D> PushOwned<T, D> {
+    /// TODO
+    pub fn new() -> (Self, Self) {
+        let zelf = Self(Rc::new(RefCell::new(None)));
+        (zelf.clone(), zelf)
+    }
+
+    /// TODO
+    pub fn set<P: Push<BundleCore<T, D>> + 'static>(&self, pusher: P) {
+        *self.0.borrow_mut() = Some(Box::new(pusher));
+    }
+}
+
+impl<T, D> Default for PushOwned<T, D> {
+    fn default() -> Self {
+        Self(Rc::new(RefCell::new(None)))
+    }
+}
+
+impl<T, D> Debug for PushOwned<T, D> {
+    fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result {
+        f.debug_struct("PushOwned").finish_non_exhaustive()
+    }
+}
+
+impl<T, D> Clone for PushOwned<T, D> {
+    fn clone(&self) -> Self {
+        Self(Rc::clone(&self.0))
+    }
+}
 
 /// Wraps a shared list of `Box<Push>` to forward pushes to. Owned by `Stream`.
 pub struct TeeCore<T, D> {
@@ -20,6 +53,17 @@ pub struct TeeCore<T, D> {
 /// [TeeCore] specialized to `Vec`-based container.
 pub type Tee<T, D> = TeeCore<T, Vec<D>>;
 
+impl<T: Data, D: Container> Push<BundleCore<T, D>> for PushOwned<T, D> {
+    #[inline]
+    fn push(&mut self, message: &mut Option<BundleCore<T, D>>) {
+        let mut pusher = self.0.borrow_mut();
+        if let Some(pusher) = pusher.as_mut() {
+            pusher.push(message);
+        }
+    }
+}
+
+
 impl<T: Data, D: Container> Push<BundleCore<T, D>> for TeeCore<T, D> {
     #[inline]
     fn push(&mut self, message: &mut Option<BundleCore<T, D>>) {

timely/src/dataflow/mod.rs

@@ -13,7 +13,7 @@
 //! });
 //! ```
 
-pub use self::stream::{StreamCore, Stream};
+pub use self::stream::{StreamCore, Stream, StreamLike, OwnedStream};
 pub use self::scopes::{Scope, ScopeParent};
 
 pub use self::operators::input::HandleCore as InputHandleCore;

timely/src/dataflow/operators/aggregation/aggregate.rs

@@ -6,6 +6,7 @@ use crate::{Data, ExchangeData};
 use crate::dataflow::{Stream, Scope};
 use crate::dataflow::operators::generic::operator::Operator;
 use crate::dataflow::channels::pact::Exchange;
+use crate::dataflow::stream::OwnedStream;
 
 /// Generic intra-timestamp aggregation
 ///
@@ -64,7 +65,7 @@ pub trait Aggregate<S: Scope, K: ExchangeData+Hash, V: ExchangeData> {
         self,
         fold: F,
         emit: E,
-        hash: H) -> Stream<S, R> where S::Timestamp: Eq;
+        hash: H) -> OwnedStream<S, Vec<R>> where S::Timestamp: Eq;
 }
 
 impl<S: Scope, K: ExchangeData+Hash+Eq, V: ExchangeData> Aggregate<S, K, V> for Stream<S, (K, V)> {
@@ -73,7 +74,7 @@ impl<S: Scope, K: ExchangeData+Hash+Eq, V: ExchangeData> Aggregate<S, K, V> for
         self,
         fold: F,
         emit: E,
-        hash: H) -> Stream<S, R> where S::Timestamp: Eq {
+        hash: H) -> OwnedStream<S, Vec<R>> where S::Timestamp: Eq {
 
         let mut aggregates = HashMap::new();
         let mut vector = Vec::new();

timely/src/dataflow/operators/aggregation/state_machine.rs

@@ -6,6 +6,7 @@ use crate::{Data, ExchangeData};
 use crate::dataflow::{Stream, Scope};
 use crate::dataflow::operators::generic::operator::Operator;
 use crate::dataflow::channels::pact::Exchange;
+use crate::dataflow::stream::OwnedStream;
 
 /// Generic state-transition machinery: each key has a state, and receives a sequence of events.
 /// Events are applied in time-order, but no other promises are made. Each state transition can
@@ -51,7 +52,7 @@ pub trait StateMachine<S: Scope, K: ExchangeData+Hash+Eq, V: ExchangeData> {
         I: IntoIterator<Item=R>,                    // type of output iterator
         F: Fn(&K, V, &mut D)->(bool, I)+'static,    // state update logic
         H: Fn(&K)->u64+'static,                     // "hash" function for keys
-    >(self, fold: F, hash: H) -> Stream<S, R> where S::Timestamp : Hash+Eq ;
+    >(self, fold: F, hash: H) -> OwnedStream<S, Vec<R>> where S::Timestamp : Hash+Eq ;
 }
 
 impl<S: Scope, K: ExchangeData+Hash+Eq, V: ExchangeData> StateMachine<S, K, V> for Stream<S, (K, V)> {
@@ -61,7 +62,7 @@ impl<S: Scope, K: ExchangeData+Hash+Eq, V: ExchangeData> StateMachine<S, K, V> f
             I: IntoIterator<Item=R>,                    // type of output iterator
             F: Fn(&K, V, &mut D)->(bool, I)+'static,    // state update logic
             H: Fn(&K)->u64+'static,                     // "hash" function for keys
-        >(self, fold: F, hash: H) -> Stream<S, R> where S::Timestamp : Hash+Eq {
+        >(self, fold: F, hash: H) -> OwnedStream<S, Vec<R>> where S::Timestamp : Hash+Eq {
 
         let mut pending: HashMap<_, Vec<(K, V)>> = HashMap::new();   // times -> (keys -> state)
         let mut states = HashMap::new();    // keys -> state

timely/src/dataflow/operators/branch.rs

@@ -4,6 +4,7 @@ use crate::dataflow::channels::pact::Pipeline;
 use crate::dataflow::operators::generic::builder_rc::OperatorBuilder;
 use crate::dataflow::{Scope, Stream, StreamCore};
 use crate::{Container, Data};
+use crate::dataflow::stream::{OwnedStream, StreamLike};
 
 /// Extension trait for `Stream`.
 pub trait Branch<S: Scope, D: Data> {
@@ -31,14 +32,14 @@ pub trait Branch<S: Scope, D: Data> {
     fn branch(
         self,
         condition: impl Fn(&S::Timestamp, &D) -> bool + 'static,
-    ) -> (Stream<S, D>, Stream<S, D>);
+    ) -> (OwnedStream<S, Vec<D>>, OwnedStream<S, Vec<D>>);
 }
 
 impl<S: Scope, D: Data> Branch<S, D> for Stream<S, D> {
     fn branch(
         self,
         condition: impl Fn(&S::Timestamp, &D) -> bool + 'static,
-    ) -> (Stream<S, D>, Stream<S, D>) {
+    ) -> (OwnedStream<S, Vec<D>>, OwnedStream<S, Vec<D>>) {
         let mut builder = OperatorBuilder::new("Branch".to_owned(), self.scope());
 
         let mut input = builder.new_input(self, Pipeline);
@@ -71,7 +72,7 @@ impl<S: Scope, D: Data> Branch<S, D> for Stream<S, D> {
 }
 
 /// Extension trait for `Stream`.
-pub trait BranchWhen<T>: Sized {
+pub trait BranchWhen<G: Scope, C: Container>: Sized {
     /// Takes one input stream and splits it into two output streams.
     /// For each time, the supplied closure is called. If it returns true,
     /// the records for that will be sent to the second returned stream, otherwise
@@ -91,11 +92,11 @@ pub trait BranchWhen<T>: Sized {
     ///     after_five.inspect(|x| println!("Times 5 and later: {:?}", x));
     /// });
     /// ```
-    fn branch_when(self, condition: impl Fn(&T) -> bool + 'static) -> (Self, Self);
+    fn branch_when(self, condition: impl Fn(&G::Timestamp) -> bool + 'static) -> (OwnedStream<G, C>, OwnedStream<G, C>);
 }
 
-impl<S: Scope, C: Container> BranchWhen<S::Timestamp> for StreamCore<S, C> {
-    fn branch_when(self, condition: impl Fn(&S::Timestamp) -> bool + 'static) -> (Self, Self) {
+impl<G: Scope, C: Container, S: StreamLike<G, C>> BranchWhen<G, C> for S {
+    fn branch_when(self, condition: impl Fn(&G::Timestamp) -> bool + 'static) -> (OwnedStream<G, C>, OwnedStream<G, C>) {
         let mut builder = OperatorBuilder::new("Branch".to_owned(), self.scope());
 
         let mut input = builder.new_input(self, Pipeline);