For a lot of us, at least for me, I think the class path hierarchy is not intuitive since I'm working inside a directory structure and it feels like that ought to be it.
Java is looking at the name of the class based on it's package path, not just the file path.
It doesn't matter if:
i'm in the local directory ./packagefoo/MainClass, or
a directory up ./packagefoo/, or
one down ./packagefoo/MainClass/foo.
The command "java packagefoo.MainClass" is running off the root %CLASSPATH% which means something significant to Java. Then from there it traverses package names, not path names like us lay coders would expect.
So if my CLASSPATH is set to %CWD%/, then "java packagefoo.MainClass" will work.
If I set the CLASSPATH to %CWD%/packagefoo/ then packagefoo.MainClass can't be found.
Always "java MainClass" means nothing, if it is a member of "package", until I rip out the java code "package packagefoo;" and move the Class File up a directory.
In fact if I change "package packagefoo;" to "package foopackage;" I have to create a subfolder under CLASSPATH/foopackage or foopackage.MainClass stops working again.
To make matters worse, Between PATH, CLASSPATH, JAVAHOME, for Windows, JDeveloper, Oracle Database, and every user name it was installed under, I think a coder trying to just get something up fast ends up brute forcing path variables and structure until something works without understanding what it means.
at least i did.