Yet another fallacious conclusion draw on the basis of a near complete misunderstanding of random, irrelevant information found on the web.

The path is always used to locate an executable when a process is started, regardless of whether it is started by the command line, the shell, or another application. Unless it's name is prefixed by the full pathname, just as in *nix.

The path is also used to locate dynamic libraries, com objects, device drivers; font libraries and a bunch of other things.

The only time associations come into play is when the system is asked to locate an appropriate executable to process a named data file. A perl (or other interpreter) source file is just a data file for the interpreter executable.

The associations mechanism is just a convenience to allow you to have the system find an appropriate executable to process a given data file. As such, it mirrors the function of the *nix shebang mechanism, but is more flexible because it works with any type of data file--graphics files; music files; video files; whatever.

You cannot go stuffing a shebang line on the front of any of those file types, hence the reason linux, OS-X, and probably others, have adopted similar mechanisms based around Mime-types.

You even misinterpreted the random page you found and linked to. In a nutshell it simply says that if you screw up configuration, things go wrong. And that's true whether that configuration is stored in Windows registry, or *nix's myriad small text files strewn about all over the file system(s). If users don't follow safe working practices (backups) when modifying configuration through either mechanism, they can break things. What's new.

Was it worth your effort?