2026-10-15 21:28 UTC
The current interview question we've been using at work is basic word wrapping. Here's a long input string. Write a function wrap40(string input) -> string that word wraps to 40 characters without breaking apart words or hyphenating. Super simple, "every line is 40 characters or less."
You can usually follow up with "how would you do 80 characters?" (trivial if they do it right, just add an int length to the function and you should only have to change one thing). It's absolutely amazing how many people cannot do this fucking simple thing. With the last girl, we watch her struggle, not even hit the "run" button in the REPL once just to test out what she had so far .. and the other two people on the panel finished the thing in like 5 minutes for fun while she struggled.
Compiler issues should be fucking trivial unless you're writing in Scala and using something with nested generic types.
Replies (1)
-
@Phobos@eientei.org 2026-10-15 21:47
@djsumdog@djsumdog.com @deprecated_ii@poa.st Early code editors and compiler output could be less than ideal, and it could be hard to figure out stuff which is trivial today. But to have that as the modern measurement for competence today is strange. Compilers and their output and tools have improved, and most would (or at least should) probably agree that's a good thing. If one would agree LLMs are a good or bad is a separate question. Yet so many "old school" programmers seems to equate the needlessly hard work of the past with skill and somehow being good.