MySQL Bug #11472, a notorious issue that developers have been half-jokingly requesting cake for fixing since 2006, has finally been resolved after 20 years. The bug became such a fixture of the database community that its resolution feels less like a bug fix and more like the end of an era.This is database archaeology. Some bugs stick around so long they stop being problems and start being landmarks. They appear in Stack Overflow answers with the weary tone of someone who's explained the workaround a thousand times. They get mentioned in conference talks. They become running jokes.Bug #11472 was one of those. The details are technical - something about subquery optimization in certain edge cases - but what made it legendary was its longevity. For two decades, developers would submit patches, argue about the correct fix, and eventually give up. The bug report comment thread became a historical document of database development philosophy.Someone even started a tradition of promising cake to whoever finally fixed it. Because when a bug lasts 20 years, what else can you do but make it a party?What this really illustrates is how technical debt accumulates in critical infrastructure. MySQL powers an enormous chunk of the internet - your bank, your email, your favorite websites. It's production code that can't break. So some bugs - especially edge cases that have known workarounds - get triaged into "we'll fix it eventually" status and then languish for decades.The developers who first reported this bug might have retired. Some were probably still in school when it was first filed. The programming languages that were trendy in 2006 are now considered legacy. And through all of it, Bug #11472 persisted.Now it's finally fixed. Cake is presumably being delivered. And somewhere, a database developer is finally closing a 20-year-old bookmark. The technology is impressive. The question is why it took two decades to fix something this old.