raboof/pekko
1
0
Fork 0
Code Issues Pull requests Projects Releases 6 Packages Wiki Activity Actions 10

!str #19005 make groupBy et al return a SubFlow

Browse source 
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.
This commit is contained in:
Roland Kuhn 2015-11-25 19:58:48 +01:00
parent 654fa41443
commit 1500d1f36d
56 changed files with 3484 additions and 720 deletions
Download patch file Download diff file Expand all files Collapse all files

62
akka-docs-dev/rst/scala/migration-guide-1.0-2.x-scala.rst
View file

@ -292,6 +292,64 @@ Update procedure
*There is no simple update procedure. The affected stages must be ported to the new ``GraphStage`` DSL manually. Please
read the* ``GraphStage`` *documentation (TODO) for details.*
GroupBy, SplitWhen and SplitAfter now return SubFlow
====================================================
Previously the ``groupBy``, ``splitWhen``, and ``splitAfter`` combinators
returned a type that included a :class:`Source` within its elements.
Transforming these substreams was only possible by nesting the respective
combinators inside a ``map`` of the outer stream. This has been made more
convenient and also safer by dropping down into transforming the substreams
instead: the return type is now a :class:`SubFlow` that does not implement the
:class:`Graph` interface and therefore only represents an unfinished
intermediate builder step. The substream mode can be ended by closing the
substreams (i.e. attaching a :class:`Sink`) or merging them back together.
Update Procedure
----------------
The transformations that were done on the substreams need to be lifted up one
level. This only works for cases where the processing topology is homogenous
for all substreams.
Example
^^^^^^^
::
Flow[Int]
// This no longer works!
.groupBy(_ % 2)
// This no longer works!
.map {
case (key, source) => source.map(_ + 3)
}
// This no longer works!
.flatten(FlattenStrategy.concat)
This is implemented now as
.. includecode:: code/docs/MigrationsScala.scala#group-flatten
Example 2
^^^^^^^^^
::
Flow[String]
// This no longer works!
.groupBy(identity)
// This no longer works!
.map {
case (key, source) => source.runFold((key, 0))((pair, word) => (key, pair._2 + 1))
}
// This no longer works!
.mapAsyncUnordered(4, identity)
This is implemented now as
.. includecode:: code/docs/MigrationsScala.scala#group-fold
Variance of Inlet and Outlet
============================
@ -468,7 +526,7 @@ should be replaced by:
.. includecode:: code/docs/MigrationsScala.scala#query-param
SynchronousFileSource and SynchronousFileSink
============================================
=============================================
Both have been replaced by ``Source.file(…)`` and ``Sink.file(…)`` due to discoverability issues
paired with names which leaked internal implementation details.
@ -559,4 +617,4 @@ Example
should be replaced by
.. includecode:: code/docs/MigrationsScala.scala#output-input-stream-source-sink
.. includecode:: code/docs/MigrationsScala.scala#output-input-stream-source-sink