Imported From: https://issues.scala-lang.org/browse/SI-6684?orig=1
Reporter: graham seed (gmseed)
Affected Versions: 2.9.2, 2.10.0-M3

@soc said:
This is weird, because the parameter names are actually correct when looking at the source file. Maybe there is some library-aux magic happening of which I'm not aware of ...

same in 2.12.4

same in 2.13.0
It is strange.

I tried searching the scaladoc website for "equals", and it's dominated by the arg0 parameter. Not a good sight, to be honest.

I thought about the parameter arg0, and I figured to do a quick search online because I think it's being generated by something really implicit. See this (https://stackoverflow.com/questions/18556400/some-function-parameters-named-arg0-arg1-etc-after-compiling-a-jar-library).

I guess the questions I have are: (1) is scaladoc generating the docs based on the compiled classes/files, and (2) is the bootstrapping/compilation process for library-aux explicitly accounting for the automatically generated argN params. I'm also asking about (2) because build.sbt only references library-aux in the variable library, and because the README inside explicitly says files residing there cannot be compiled by normal means.

Note that at compile time the parameter of equals is called x$1, which is also some default name for parameters whose names are not present in the class file.

What I find the weirdest is that the Scaladoc uses arg0 instead of x$1, which suggests that it doesn't just take the names from the method symbol in the compiler, but it doesn't use the names from the source code either.

@Jasper-M Where should I start looking? I'm definitely a beginner with all the compiler stuff.

I don't know anything about scaladoc, but the class symbols for Any and AnyRef, and the symbols for their methods are generated programmatically in scala.reflect.internal.Definitions. If you look a bit further you find that it uses the method Symbol#newSyntheticValueParams which explicitly generates those nondescriptive x$1 names. Another difficulty is that some AnyRef method symbols are not generated but read (also in Definitions) from java.lang.Object whose classfile doesn't store parameter names.

And I have to reiterate that I'm not sure this is actually related to the scaladoc-specific issue. It could definitely help to fix a related named arguments issue, which I don't know the issue number of, if any exists.