@SuppressModernizer not applicable to type use

[delanym]

I'm unable to suppress the StringBuffer to StringBuilder recommentation:

StringBuffer sb = new @SuppressModernizer StringBuffer();

If I say

@SuppressModernizer
StringBuffer sb = new StringBuffer();

I get

'@SuppressModernizer' not applicable to local variable

[delanym]

I could maybe use the exclusions tag, but what the heck is a javap format: java/lang/String.getBytes:(Ljava/lang/String;)[B supposed to mean I dont know. Can you include a couple of examples? Presumably most people using this plugin have no knowledge of bytecode internals.

[gaul]

@stevegutz could you add support for @SupressModernizer on local variables? I believe that Java 8 and later allow this. Adding support for constructors was easy via #126.

@delanym this mangled name is unusual but unambiguous. If you run your class file through javap -c you should be able to grep for the method name and see the type mangling. If you have any better suggestions please open an issue so we can improve this.

[stevie400]

Hi, my org is using error-prone instead of modernizer so I'm not going to be contributing any more.

[rovarga]

@gaul Perhaps the first step is to add ElementType.TYPE_USE, which should be easy if annotations were bumped to 1.8 or if it were a multi-release jar. The second part is dealing with the annotation at detection point, I think: a TYPE_USE occurrence needs to be propagated to the corresponding variable allocation, which again would require JDK8+ (9+ in case of multi-release jar) as the runtime. That would satisfy everyone at Java 11+ (mulit-release, which works for Java 9+). Java 8 users would be left off no worse than they are now.

[gaul]

I'm not too sure what the current minimum version is but JDK 6 used to work. I am happy to declare Java 8 the new minimum so Modernizer can access ElementType.TYPE_USE. Can you propose a PR for this?

[gaul]

I would like the next 2.6.0 release to include this feature. @delanym @rovarga could you look into this?

[rovarga]

Yes, PR coming out in a few minutes.

[rovarga]

171 , but it also touches definitions -- sun.misc.BASE64Decoder/Encoder no longer exist.

[delanym]

Missed the release :(

[gaul]

Just to be clear there is no fix provided. There is an open linked PR with no explanation for what it does.

[gaul]

I still don't understand if this is possible to fix. https://docs.oracle.com/javase/specs/jls/se8/html/jls-9.html#jls-9.6.4.2 says:

An annotation on a local variable declaration is never retained in the binary representation.