Repository navigation
Removing unused imports leaves extra blank line after package statement, isn't idempotent #598
Description
I have a class that if I format it twice with google java formatter it will produce 2 different results. This is not what I would expect - I would think that you can format a file any number of times and google java formatter should always produce the exact same result. Is my assumption wrong ?
Here is an example:
original class:
$ cat inital_class.java
package com.sun.something;
import com.x;
import com.y;
import com.z;
public class MockedLiveServiceExecutionContext {
}
$
If I then format it I get this:
$ java -jar ./google-java-format-1.9-all-deps.jar --replace inital_class.java
$ cat inital_class.java
package com.sun.something;
public class MockedLiveServiceExecutionContext {}
$ cat inital_class.java | cksum
42955951 79
$
And if I format it again I get this (it's removed an extra space):
$ java -jar ./google-java-format-1.9-all-deps.jar --replace inital_class.java
$ cat inital_class.java
package com.sun.something;
public class MockedLiveServiceExecutionContext {}
$ cat inital_class.java | cksum
2486624596 78
$
Reacted by Kevin Bourrillion, Matthias Schlaipfer and Ulysse Mavrocordatos
Activity
[-]edge case where google java formatter not Idempotence[/-][+]edge case where google java formatter is not idempotence[/+][-]edge case where google java formatter is not idempotence[/-][+]edge case where google java formatter is not idempotent[/+]kevinb9n commentedon Jun 14, 2021
Thanks for finding this. We would agree that any non-idempotence is a bug. Since it at least reaches a fixed point quickly (it's like "idempotence, once removed") it might not be a high-priority fix, but we're glad to keep this open until then.
[-]edge case where google java formatter is not idempotent[/-][+]Removing unused imports leaves extra blank line after package statement, isn't idempotent[/+]cushon commentedon Oct 29, 2026
Fixed in 633ce6b and f7eeb77, which will be included in the next release.