Why are packages labled com.person.mypackage rather than just person.mypackage
They used DNS domains as a namespace to keep one company’s work from colliding with another.
To expand on this, it’s the naming convention that Java defines.
Package names are written in all lower case to avoid conflict with the names of classes or interfaces.
Companies use their reversed Internet domain name to begin their package names—for example, com.example.mypackage for a package named mypackage created by a programmer at example.com.https://docs.oracle.com/javase/tutorial/java/package/namingpkgs.html
Further it isn’t always
com, you’ll often seenet,org, and rarely evenioshow up as the base package. All that said with the introduction of modules and some of the other recent changes I’m not sure how much longer we’re going to see the reversed DNS naming system stick around.I’ve quite often seen people shoehorn their Github account into this format. com.github.username.projectname is quite a common sight for Flatpak packages. Reverse DNS naming assumes every package is created by a company while nowadays a large portion of popular packages were created by individuals.
It doesn’t check for an actual DNS entry though, right? Would it be possible to do
person.developer.app?Correct, there’s no actual validation done. You can literally make the package anything you want, you don’t even need to follow the convention if you don’t want to, although it is of course considered best practice to do so. The entire reason it exists is just as a convenient way to namespace Java classes without worrying about collisions. I personally think using
com.github.username.projectnameis a perfectly fine way to use the system and is completely within the spirit of it. Likewise if you wanted to useperson.developer.appyou could, although that’s a little weirder. You might be better off using something likedev.person.app, although you’d need to be careful you don’t accidentally step on someones toes if someone actually has the registration forperson.dev.
I addition to the answers others have given, consider the large centralized package repositories for software development: npm, PyPI, crates.io, etc. If you frequently publish packages to these indexes, you often will find name collisions and are forced to come up with clever names. In the case of PyPI specifically, it is even worse because while your pip package might be uniquely named, the actual importable python package may not be, and you end up where two pip packages clobber each other. Java’s system is verbose but it is simple and solved this problem decades ago.
I always thought about this, but I’ve never bothered to search for the answer. So thanks for asking!
