A SubFlow (or SubSource) is not a Graph, it is an unfinished builder
that accepts transformations. This allows us to capture the substreams’
transformations before materializing the flow, which will be very
helpful in fully fusing all operators.
Another change is that groupBy now requires a maxSubstreams parameter in
order to bound its resource usage. In exchange the matching merge can be
unbounded. This trades silent deadlock for explicit stream failure.
This commit also changes all uses of Predef.identity to use `conforms`
and removes the HTTP impl.util.identityFunc.
main work by @drewhk with contributions from @2m and @rkuhn
This work uncovered many well-hidden bugs in existing Stages, in
particular StatefulStage. These were hidden by the behavior of
OneBoundedInterpreter that normally behaves more orderly than it
guarantees in general, especially with respect to the timeliness of
delivery of upstream termination signals; the bugs were then that
internal state was not flushed when onComplete arrived “too early”.
* also make factories more consistent by only offering
FlowGraph.create()
* also remove secondary (edge-based) FlowGraph.Builder DSL
* also improve naming for conversions from Graph to
Source/Flow/BidiFlow/Sink
- replace all occurrences with equivalent GraphStage implementations
This commit introduces a mini-DSL for GraphStage that allows emitting or
reading multiple elements to/from a port with one statement, installing
stateful handlers on the port to make it work. The emitting side allows
stacked continuations, meaning that while an emit() is ongoing (waiting
for demand) another one can be added to the queue; this allows
convenient formualation of merge-type stages.
This necessitates the removal of method overloading in the Java Graph
DSL: the to() and via() methods were not otherwise resolved correctly by
javac, leading to incomprehensible error messages. The new approach is
to offer just one way of doing things which is a bit more verbose but
should be easier to read and learn. In this vein auto-importing while
using the DSL is also gone for Java—not sure about Scala yet.
- explicitly provide Unit values and place parens around tuple creation
- remove structural type usage in TestUtils
- fix Java double-casts
- use unused Java values by asserting their non-nullness
- work around inability to place case class in trait (scripted test)
The remaining warnings about using private types in public methods are
bogus as reported in https://issues.scala-lang.org/browse/SI-9490.
+ with javadsl
+ allows configuring log levels
+ allows turning off logging of certain actions completely
+ cookbook adjusted to show this instead of manual PushStage
- PENDING: preStart based impl will be faster, coming soon
