Hacker News Reader

Comment by ctoth | Hacker News Reader

So glad somebody finally said this. I've noticed this dynamic for years and it really is sad to watch. I remember reading over the community reviewing a large patchset someone had contributed a couple months ago and the amount of negativity was remarkable.

View on HN

I had one conversation with Guido at Dropbox where I wanted to get a dependency updated so that I could write quickcheck tests.

It was like pulling teeth - he didn't know what property testing was ("some haskell thing?" - his words) and so he wouldn't update it ("are you implying that how Dropbox writes tests today isn't good enough?" - his words), so I couldn't write generated tests.

Sort of ironically, property testing was used elsewhere in the company to great success, and in fact we were all told the scary story of the 1 hour where empty passwords let anyone log into any account - a bug that a proeprty test is very well designed for.

I walked away pretty embarrassed for him and decided I wasn't interested in further interaction.

Isn't it really a "haskell thing", though? Generated testing of properties seems a bit overkill, but I might be mistaken about how that actually works.

quickcheck was initially a haskell project, but it's a little absurd to reduce proeprty testing down like that.

I also wouldn't call it overkill. It's trivial to do generated tests.

Would you mind telling me exactly which patchset you are referring to?

Not OP but probably this one: https://news.ycombinator.com/item?id=28896367.

Reply on news.ycombinator.com