News for hackers. Stuff that matters.
Re:Interview Programming Problems Done Right by mikehoward on Friday January 06, 2012 @12:14AM (#3432074)
Problems on a whiteboard don't test programming skills simply because the whiteboard isn't executable. You don't get the feedback that REPL gives you, so you can't build up your solution - and understanding of the problem - iteratively.

Seems to me that holding a code retreat and pairing with interviewee's would be more realistic. It would eliminate at least some of the pointless stress of interviewing and would get closer to a real programming environment.

It's not like nobody has enough computers so you have to use whiteboards.

Re:Interview Programming Problems Done Right by gcp on Friday January 06, 2012 @04:16AM (#3432626)
Problems on a whiteboard don't test programming skills simply because the whiteboard isn't executable. You don't get the feedback that REPL gives you, so you can't build up your solution - and understanding of the problem - iteratively.

Strange argument. What are people who write in compiled languages supposed to do? Particularly if their code is part of a product with a long build cycle?

It's not like nobody has enough computers so you have to use whiteboards.

On this I agree. We switched from whiteboards to giving the candidate a laptop, their editor of choice, and access to a compiler and the manpages and this was a major improvement.

Re:Interview Programming Problems Done Right by mikehoward on Wednesday January 18, 2012 @08:52PM (#3482853)
look up doc - edit - compile - test - repeat is pretty close to interpreted language repl cycles. But then I started on punch cards, so maybe my perspective is different. [one to three compile & test cycles/day makes anything interactive look good]

Appreciate the comment though. Gives me food for thought.