An integer formula for Fibonacci numbers (2015) | Hacker News Reader

An integer formula for Fibonacci numbers (2015)

blog.paulhankin.net

93 pointsslbenfica19 comments

Now you're really gonna' piss the interviewers off!

http://joelgrus.com/2016/05/23/fizz-buzz-in-tensorflow/

Neat text. I'd be really impressed if I were to interview him, but then again: I'd never ask someone to write a fizzbuzz... hmmm...

Edit: I actually tried a FizzBuzz myself and I actually like it. I get why people were to ask it at interviews since it does require you to think for a second (if you haven't done it before) so you can easily filter out people who can't think

Indeed, what's the point to ask someone to write a fizzbuzz when it's already been done to the highest of professional enterprise standards! [1]

[1] https://github.com/EnterpriseQualityCoding/FizzBuzzEnterpris...

Highest professional enterprise standards my arse. Where's the message queues? Why isn't there a Fizz microservice and a Buzz microservice that live behind an API Gateway and receive messages from an event driven architecture?

I'm gonna need a write up. I've been looking at the code for 10 or 15 minutes and I still haven't found "fizz", "buzz", or the number 100 anywhere in the code. It's amazing.

I have no idea either, but those are "constants", and so it is only logical that they belong in com.seriouscompany.business.java.fizzbuzz.packagenamingpackage.impl.Constants (https://github.com/EnterpriseQualityCoding/FizzBuzzEnterpris...).

I think it’s really shortsighted to hardcode them statically into constants. You’d have to build a completely new binary should the requirements someday change! Obviously they belong behind a configuration interface, an instance of which can then be dependency injected to places that need them. In this way you can easily supply them from whatever source you like, such as a properties file, XML document, a database or even a web service.

As it is now, 2/10 would not pass review.

Sounds like Spring ConfigurationProperties https://docs.spring.io/spring-boot/docs/current/reference/ht...

I'm utterly unsurprised.

Found it!!

https://github.com/EnterpriseQualityCoding/FizzBuzzEnterpris...

IsEvenlyDivisibleStrategyFactory.java ... now I remember why I hate java.

Needs more AWS Lambda, this is outdated.

It doesn't require you to think - it makes you think there is an elegant solution but there isn't really. I suspect you've done it wrong.

Many many years ago I did a project building a functional programming language that macro expanded into lambda calculus that was then compiled down to SK combinators for evaluation. This was entertainingly inefficient.

I'm tempted to resurrect the code (I still have a printout) and use it to implement Fizz Buzz....

You could take a look at what this guy[1] did. It was a great read for me as a real beginner in lambda calculus.

[1]: https://www.jtolio.com/2017/03/whiteboard-problems-in-pure-l...

Another article on the HN front page also talks about computing Fibonacci numbers quickly.

"Automatic Algorithms Optimization via Fast Matrix Exponentiation (2015)" ; https://news.ycombinator.com/item?id=17592359 ; https://kukuruku.co/post/automatic-algorithms-optimization-v...

Previous discussion: https://news.ycombinator.com/item?id=11560122 and Paul Hankin has some more thoughts on the subject: https://blog.paulhankin.net/fibonacci2/

Discussion on Reddit: https://www.reddit.com/r/math/comments/910dlj/an_integer_for...

Still looking at the formula and puffing my cheeks out in disbelief. Every time I read it again it seems weirder. Bitwise and?

Reply on news.ycombinator.com