If you're scoffing at this, I want you to do an experiment. Imagine yourself at 26, when you just knew enough to be dangerous. Or as the person you're currently mentoring/would like to mentor. Now go into your application code, set a breakpoint on one of the moderately complex `logger.info()` calls, run your application and step into log4j. All the way down, until it writes to disk.

If you don't say "what the fuck" out loud at least once, you're a soulless monster and should go live in a cave.

If your application has a pretty good architecture, you may well discover that the code leading up to the `logger.info()` call is substantially less complex than the code below it. Just to write some text to the end of a single-writer file.

It chaps my ass that log4js, which claims specifically not to be a javascript port of log4j, has the same pattern of layers and weird indirection (some of which are avoided by monkey patching, which is in many respects even worse from a legibility standpoint, and absolutely from a discoverability standpoint). Much of this is chasing after not implementing the same 1 line function 6 times, which is made weirder by the fact that you have to write individual unit tests for each of the log levels, so you've already written 95% of the duplicated code.

Logging frameworks are a nearly canonical example of DRY as an antipattern.

Every new project I see defaults to slf4j/logback.