Improve SimpleJsonEncoder.escapeString memory usage #61
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
|
I also created and ran a Java microbenchmark: Before: After: Is notable the reduced GC count and time, and increased throughput, almost 3x |
osiegmar
requested changes
May 7, 2021
| } | ||
| if (ch < ' ') { |
There was a problem hiding this comment.
Why don't you keep this if in the switch (default) block?
There was a problem hiding this comment.
The code looked simpler that way, but I didn't realize that it would add an unnecessary if condition check when the current char is an escaped char. Moved the if back to the default block
| escaped = escapeCharacter(ch); | ||
| } | ||
| if (escaped != null) { | ||
| writer.append(str.substring(lastCopyIndex, currentIndex)); |
There was a problem hiding this comment.
As this method already gets the char array representation of str (via str.toCharArray()), I assume that using java.io.Writer#write(char[], int, int) would be a lot faster than slicing the string. Could you check this?
There was a problem hiding this comment.
Changed to Writer.write instead of append, and it performed way better than the first changes. Also performed better than the upstream version of this code, I will put the reports on the comments bellow
| currentIndex++; | ||
| } | ||
| if (lastCopyIndex != currentIndex) { | ||
| writer.append(str.substring(lastCopyIndex)); |
There was a problem hiding this comment.
...same here java.io.Writer#write(char[], int, int) is probably way faster.
|
See the summary of the benchmark results between the upstream version and before this changes: Upstream: After: testEncode -> 25% throughtput gain My understanding is that with the current changes, there are gains when there are few characters to escape, but there are losses when encoding strings with more characters to escape. I'm going to think how to re-work those changes. I also uploaded the benchmark code on a new git repo: https://github.com/fabiojb/logbook-gelf-benchmark |
fabiojb
changed the title
osiegmar
reviewed
May 11, 2021
| int lastCopyIndex = 0; | ||
| char[] chars = str.toCharArray(); | ||
| for (final char ch : chars) { | ||
| String escaped = null; | ||
| switch (ch) { |
There was a problem hiding this comment.
It might be beneficial to refactor the switch block to a private static method and call it via final String escaped = escapeIfNeeded(ch);. Another possible optimization would be to have private static final char attributes for all the escape sequences to test for.
There was a problem hiding this comment.
I'm going to work on a way to escape the string only if is needed.
I did some tests putting some values on static final variables, but it didn't show a significant difference in the performance tests
|
@osiegmar I have submitted two new changes, and finished the improvements I had in mind. First 90fb46a, reduced memory allocation: Second, 53ca94b is a more verbose change, that escapes characters only when it's needed. It did improve the throughput: To better understand the improvements, you can compare this results against the Upstream results in the previous comment. |
fabiojb
changed the title
osiegmar
requested changes
May 17, 2021
fabiojb
force-pushed
the
json-encoder-performance
branch
2 times, most recently
from
d1cc3a0 to
5d512cd
Compare
|
No problem, I understand your point. I removed the last commit, but can you still review the commit ca7259f that improves memory usage? |
fabiojb
force-pushed
the
json-encoder-performance
branch
from
5d512cd to
ca7259f
Compare
osiegmar
requested changes
May 18, 2021
| @@ -107,7 +107,8 @@ private void appendKey(final String key) throws IOException { | |||
| */ | |||
| @SuppressWarnings("checkstyle:cyclomaticcomplexity") | |||
| private void escapeString(final String str) throws IOException { | |||
| for (final char ch : str.toCharArray()) { | |||
| for (int i = 0; i < str.length(); i++) { | |||
| char ch = str.charAt(i); | |||
There was a problem hiding this comment.
Please add a final in front of char ch and I can approve it.
Iterate over the current String internal values instead of making a copy of its content with String.toCharArray
fabiojb
force-pushed
the
json-encoder-performance
branch
from
ca7259f to
ee36fad
Compare
fabiojb
changed the title
|
I'll be part of https://github.com/osiegmar/logback-gelf/milestone/1 which is to be released when the test coverage is (greatly) improved. Unfortunately I hadn't had enough time for that and cannot promise a specific date. |
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.
write String slices instead of single chars to the Writer. Iterate over the string to be escaped and write a new slice every time a char needs to be escaped.