-
Notifications
You must be signed in to change notification settings - Fork 1k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Improve performance of NamingConvention#toSnakeCase #5271
base: main
Are you sure you want to change the base?
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Improving the performance of this would be a nice enhancement for many users, but we need to be careful as the impact of issues here could be large.
@@ -57,7 +57,7 @@ public String tagKey(String key) { | |||
} | |||
|
|||
private String toSnakeCase(String value) { | |||
return Arrays.stream(value.split("\\.")).filter(Objects::nonNull).collect(Collectors.joining("_")); | |||
return value.replace(".", "_"); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
It's subtle, but this does not have the same behavior as the previous code. See, for example, the following test on the main
branch (i.e. without this code change):
@Test
void snakeCaseReplace() {
String name = "a.Name.with.Words.";
String expectedConventionName = "a_Name_with_Words";
String actualConventionName;
actualConventionName = NamingConvention.snakeCase.name(name, Meter.Type.COUNTER);
assertThat(actualConventionName).isEqualTo(expectedConventionName);
actualConventionName = name.replace(".", "_");
assertThat(actualConventionName).isEqualTo(expectedConventionName);
}
The second assertion fails:
org.opentest4j.AssertionFailedError:
expected: "a_Name_with_Words"
but was: "a_Name_with_Words_"
It's unfortunate we don't have existing tests to cover a more complete set of expected input and output, but it would be risky to change this behavior and cause potential regressions for users.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
If #5288 increases performance and is more consistent with module design, I'm fine and this PR can be closed.
According to "subtle behavior change", it should be documented how metric name is changed for Prometheus. Then it will be possible to say if removing leading and trailing dots is part of contract or random side effect
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I think some benchmarks (JMH) would be also nice to prove that this is in fact an improvement. Regarding #5288, this might not be the root cause since this line is there for 7 years and as far as I understand the issue is new.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Here you are:
@Benchmark
public void snakeStream(Blackhole blackhole) {
String value = "http.client.requests.active.seconds.max";
String snak = Arrays.stream(value.split("\\.")).filter(Objects::nonNull).collect(Collectors.joining("_"));
blackhole.consume(snak);
}
@Benchmark
public void snakeSimple(Blackhole blackhole) {
String value = "http.client.requests.active.seconds.max";
String snak = value.replace(".", "_");
blackhole.consume(snak);
}
Benchmark Mode Cnt Score Error Units
SnakeCase.snakeSimple thrpt 25 49690145,331 ± 3281631,270 ops/s
SnakeCase.snakeStream thrpt 25 3717968,300 ± 101394,596 ops/s
No description provided.