I interviewed at Microsoft and it was about 10 hours of this stuff and I loved it. Had a great time, and I think I did pretty well. One guy wanted me to write a shuffling program. So I came up with one which I later discovered was Fisher-Yates and optimal. But he kept saying "how can it be faster" so I'm not sure what I was missing. Most of the other stuff was pretty fun and straightforward - pointer stuff, string copying, synchronization. It felt like if you had gone through Sedgewick's algorithm book you'd be set for the most part.
Unfortunately by 7pm or so I was sorta exhausted (didn't sleep the night before 'cause I was so excited). I met the hiring manager, and he asked something fun like make a ring buffer and also write a non-recursive inorder traversal function of a binary tree. I apparently had some off-by-one bug in the ring buffer code, and then I just blanked out on the traversal question. No hire :\.
(OTOH, that team wasn't sure if they were going to be using VB6, or C++, or maybe .NET, but probably not because they wanted to ship with Office and Office severely limits dependencies. So maybe it was for the better.)
Unfortunately by 7pm or so I was sorta exhausted (didn't sleep the night before 'cause I was so excited). I met the hiring manager, and he asked something fun like make a ring buffer and also write a non-recursive inorder traversal function of a binary tree. I apparently had some off-by-one bug in the ring buffer code, and then I just blanked out on the traversal question. No hire :\.
(OTOH, that team wasn't sure if they were going to be using VB6, or C++, or maybe .NET, but probably not because they wanted to ship with Office and Office severely limits dependencies. So maybe it was for the better.)