Add Publisher.fromInputStream(InputStream, ByteArrayMapper)
#2989
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Motivation: Existing `Publisher.fromInputStream(InputStream)` contract has a side effect of an extra allocation in case the pre-allocated byte-array was not completely full of data. Modifications: - Add `Publisher.fromInputStream(InputStream, ByteArrayMapper)` that allows users to decide how to use the buffer region with data; - Deprecate pre-existing `Publisher.fromInputStream(InputStream)` and `Publisher.fromInputStream(InputStream, int)` overloads; - Update all existing use-cases to use `BufferAllocator.wrap` as a method reference as a `ByteArrayMapper`; Result: No impact on throughput and latency for `BlockingStreamingHttpClient` requests with `InputStream` payload, but significant reduction in memory allocations and number of GC runs per benchmark: Allocations: ~2Gb/s -> ~1.4Gb/s Young Collection GC Count: 29 -> 23 Alloc Outside TLABs: 7.33 GiB -> 5.04 GiB
idelpivnitskiy
commented
Jun 26, 2024
| * @param <T> Type of the result of this mapper | ||
| */ | ||
| @FunctionalInterface | ||
| public interface ByteArrayMapper<T> { |
There was a problem hiding this comment.
We need this interface bcz concurrent-api doesn't have a dependency on buffer-api module.
Eventually, I think it's worth it. We can consolidate configuration parameters for how to read data from InputStream as default methods on this interface instead of adding more and more overloads for Publisher.fromInputStream. Publisher is already quite a lengthy class.
bryce-anderson
approved these changes
Jun 26, 2024
servicetalk-concurrent-api/src/main/java/io/servicetalk/concurrent/api/ByteArrayMapper.java
Outdated
Show resolved
Hide resolved
bryce-anderson
approved these changes
Jun 26, 2024
daschl
reviewed
Jun 27, 2024
servicetalk-concurrent-api/src/main/java/io/servicetalk/concurrent/api/ByteArrayMapper.java
Outdated
Show resolved
Hide resolved
| * @param <T> Type of the result of this mapper | ||
| */ | ||
| @FunctionalInterface | ||
| public interface ByteArrayMapper<T> { |
...alk-concurrent-api/src/main/java/io/servicetalk/concurrent/api/FromInputStreamPublisher.java
Show resolved
Hide resolved
daschl
approved these changes
Jun 28, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Motivation:
Existing
Publisher.fromInputStream(InputStream)contract has a side effect of an extra allocation in case the pre-allocated byte-array was not completely full of data.Modifications:
Publisher.fromInputStream(InputStream, ByteArrayMapper)that allows users to decide how to use the buffer region with data;Publisher.fromInputStream(InputStream)andPublisher.fromInputStream(InputStream, int)overloads;BufferAllocator.wrapas a method reference as aByteArrayMapper;Result:
No impact on throughput and latency for
BlockingStreamingHttpClientrequests withInputStreampayload (8Kb, 16Kb, 24Kb), but significant reduction in memory allocations and number of GC runs per benchmark:Allocations: ~2Gb/s -> ~1.4Gb/s
Young Collection GC Count: 29 -> 23
Alloc Outside TLABs: 7.33 GiB -> 5.04 GiB