News for hackers. Stuff that matters.

A comment on On Marketing Haskell.

Re:On Marketing Haskell by dmix on Saturday May 30, 2020 @01:23PM (#23362822)
One of the best ways to get adoption is to have people take initial risks and build real world serious product with a serious team/company backing behind it.

Erlang got a big boost when it came out WhatsApp scaled it up with a billion users, with tiny team and only a few servers.

React/React Native got boosted by Instagram adopting it to build their production mobile app.

Rails had 37signals/Basecamp and Github/Shopify/early Twitter/etc.

Haskell needs some more famous commercial "killer-app" examples for real businesses and real teams. And yes I'm very familiar with the current selection of popular Haskell examples.

Currently I only ever hear Haskell (and Ocaml for that matter) being used by engineers for backend engineery things (parsing programming languages, fighting spam, tooling) mostly for OSS tools.

On a positive note, PureScript is trying to position itself as a practical Haskell and making a bit of success. One helpful reason is they have a nice app/company leading the way as a production app with Lumi [1], with their founder playing a key role on the language/ecosystem dev side as well: https://www.lumi.dev/blog/purescript-and-haskell-at-lumi

This is what Haskell needs more of - and critical a company promoting/blogging about their progress using Haskell as well as contributing back into the community.

Another big one is simply a sort of 'starter kit' which guides you with best practices on how to architect a serious Haskell app (ie, using STM, free, etc). React has plenty of starter kits. For Rust CLI apps there is a great starter kit https://github.com/iqlusioninc/abscissa

I'd love to see a similar starting point micro-framework for starting a new Haskell app that makes a bunch of importation decisions for you and guides the structure of the project.

[ Parent | on HN ]
Re:On Marketing Haskell by centimeter on Saturday May 30, 2020 @01:30PM (#23362884)
I’m aware of 3-5 large companies (the number depending on what you define as “large”) that are using Haskell for big, serious projects, but all but one of them is in the defense or banking industries. Companies in these industries are usually not interested in sharing their experiences or engineering techniques, so it feels very quiet compared to when some hip startup starts (loudly) marketing some technology they’re using.
[ Parent | on HN ]
Re:On Marketing Haskell by narrationbox on Saturday May 30, 2020 @01:27PM (#23362859)
There is Hasura, they are probably the most famous commercial Haskell app right now (in terms of brand recognition).

https://hasura.io

[ Parent | on HN ]
Re:On Marketing Haskell by dmix on Saturday May 30, 2020 @03:49PM (#23364024)
That fits into the tooling category but it is a serious company.

I found one blog post mentioning Haskell but without much technical details:

https://hasura.io/blog/from-zero-to-hipster-haskell-in-produ...

I hope they write more..

[ Parent | on HN ]
Re:On Marketing Haskell by chrischen on Saturday May 30, 2020 @02:42PM (#23363512)
Ocaml (via ReasonML syntax) accounts for over 50% of Messenger.com. frontend code.

https://reasonml.github.io/blog/2017/09/08/messenger-50-reas...

I’ve personally found starting adoption by converting pure domain code first to be the best approach, since it does not have to interact with legacy code.

[ Parent | on HN ]
Re:On Marketing Haskell by sigrlami on Saturday May 30, 2020 @08:05PM (#23365988)
There are more than 200 companies using Haskell in a different levels, check out https://haskellcosm.com
[ Parent | on HN ]
Re:On Marketing Haskell by tayo42 on Saturday May 30, 2020 @02:55PM (#23363626)
a tagent, is erlang and maybe elixir worth spending time learning if Ive spent time writing haskell and scala?
[ Parent | on HN ]
Re:On Marketing Haskell by dmix on Saturday May 30, 2020 @03:53PM (#23364074)
Yes, I loved learning Erlang (and later Elixir was a breeze). It's a good experience in learning how to build apps using the Actor model (they predated it before it was called Actors) and building resilient, multi-process based systems where failure is part of the design and fully accounted for.

It also has a really small and simple syntax and wonderful pattern matching heavy approach to code, which is something I've took with me to other languages (and a feature sorely missing in most major languages).

OTP is a bit of a beast and feels overwhelming at first, for example when you download an HTTP server to run your app you actually "run" the supervisor which spawns a bunch of processes. And this is all within your own application which is a group of isolated processes.

I'd personally use Erlang/Elixir for everything if it had strong typing. And I still might anyway for my next project.

[ Parent | on HN ]