C++ says “We have try... finally at home”

HN title mangling and meme confusion

• The original blog title referenced the meme “we have X at home” and specifically “try…finally”; the HN-submitted title dropped “finally,” changing the meaning.
• Commenters argue this misrepresents the post (sounds like a criticism of try, not a discussion of finally-like behavior) and violates HN’s “don’t editorialize titles” norm.
• Some defend HN’s automatic de-clickbaiting as mostly effective; others call it a persistent “bug” that often distorts titles and depends too much on users emailing mods.
• There is debate over whether using memes in technical titles is wise, since non‑cultural insiders may miss the joke entirely.

Destructors, RAII, and “finally”

• One camp: C++ destructors and RAII are strictly superior to finally for resource management: you write cleanup logic once in the destructor, not in every finally block, and nested try/finally indentation disappears.
• Opposing view: finally and destructors solve different problems—destructors encode ownership; finally is function/scope‑level cleanup for arbitrary actions, including things you can’t or shouldn’t model as an object.
• Critics of “just write a class” say it’s overkill for a one‑line cleanup; defenders respond that serious resources should already be wrapped in RAII types, and ad‑hoc scope guards or library helpers cover the rest.

Defer, context managers, and other languages

• Swift and Go’s defer are praised as more general “run at scope exit” constructs that don’t add nesting and place setup and teardown near each other in the code.
• Others prefer Python‑style context managers / C# using / Java try‑with‑resources, arguing they make ownership explicit, at the cost of extra syntax and indentation.
• There’s discussion of macro‑ and library‑based scope guards in C++ and Rust that approximate defer.

Exceptions, destructors, and finally footguns

• Several comments focus on error semantics: many languages let exceptions from finally override the original, which is considered a design mistake that hides the root cause.
• C++’s rule that an uncaught exception thrown during stack unwinding terminates the program is criticized as harsh; others say it’s necessary and that destructors “shouldn’t throw” in practice.
• Java‑style finally is shown to allow pathologically confusing control flow (e.g., returning from finally and discarding a thrown error).