Hacker News Reader

Comment by aaronbrethorst | Hacker News Reader

Costs a lot of money to build your own programming language and then throw it out. https://blog.asana.com/2017/08/performance-asana-app-rewrite...

View on HN

"Our founding engineers had learned from their experience working at Google and Facebook that to ensure a performant and stable application, they would need to build their own framework"

Wow. Interesting.

I didn't believe that was an actual quote from the article until I read it. It sounds like a joke and I can't imagine anyone reasonable would say that.

There are definitely good reasons to build your own framework but performant and stable applications isn't one of them. Usually it's a solution looking for a problem.

In fairness the decision was made in 2009. I still don’t think it was the right decision but it feels a lot more understandable to have made that choice when looking at the frameworks available in 09 rather than 2020.

Was this a front end web framework or back end?

I never worked at Asana but I heard it was both, it had code sharing between client and server, some server-side javascript before node.js (I think it was based on JSC) and something similar to isomorphic javascript before this term was coined by Airbnb. If you look at Meteor, a lot of similar ideas migrated there.

By this logic nobody would ever build a framework designed around performance and stability?

Stability is the result of smoothing out bugs the hard way. A new framework has bugs, you just haven't found them yet.

What? Facebook built react to solve these problems.

When they built Luna stuff like react didn't exist.

Investing in developer tooling including frameworks is a fine use of time and often pays off hugely. There wasn't a better alternative 11 years ago.

weird. Not sure how they would have been hired at Google or FB with this mindset.

Dustin Moskovitz, one of the "founding engineers", was a co-founder of Facebook.

just use nodejs

Keep in mind, Asana was founded in 2008.

Luna had some of the same concepts underpinning React, about 5 years earlier.

At Google or Facebook that makes sense. If you encounter much larger scale than pretty much any other company, it's hard to find libraries that work for you. So the observation was correct but they should've concluded that this is just a result of scale and complexity, not a general strategy.

That was in 2009 apparently. Maybe it made sense then.

Luna wasn’t a programming language, it was a framework. The link you yourself provided equates it to “React, RxJS and Firebase” and states the closest OSS equivalent as Meteor (developed by ex-Asana employees).

Sorry, my bad. Here's the blog post where they announced LunaScript, the ill-fated programming language that they built and then tossed.

https://blog.asana.com/2010/02/lunascript-our-in-house-langu...

Reply on news.ycombinator.com