3 ms·

are people without a degree in type systems even allowed to express their opinion on dynamic vs static? because if their opinion is automatically viewed as infe

are people without a degree in type systems even allowed to express their opinion on dynamic vs static? because if their opinion is automatically viewed as inferior whenever a person with a degree chimes into a discussion, isn't dynamic typing stuck in the limbo of not having competent defenders and the only path for a defender to become competent is to spend years of their lives learning the opposite thing? i think (as a person without a degree in type systems) that relative success of Erlang and Clojure are a testament that pragmatic approach works, and that is not in any way evidence that static typing is inherently bad or can't work.

dgllghr 9y ago
Erlang & Elixir have a very interesting approach to typing. The support for pattern matching in function definitions provides a kind of structural typing for functions. And they both embrace typespecs for type analysis through a separate program analyzer. So a lot of the benefits of static typing are there without the static types.
- keymone 9y ago
  clojure has destructuring, which isn't quite pattern matching but it does help to detect wrongly shaped inputs. also dialyzer is amazing and i wonder if clojure.spec can be used like that.
  - DigitalJack 9y ago
    clojure has pattern matching too. As a library. Much of the interesting functionality of clojure comes from libraries. (core.async, core.spec, core.match, etc..) The matching library is good. I wouldn't use it in a tight performance loop, but as a productivity booster it's very nice.
cutler 9y ago
Relative success? In what sense? As far as the jobs market is concerned Erlang/Elixir and Clojure have failed to make an impression. Ok, Wallmart, Bleacher Report etc. but search Indeed.com by job title and you'll get what I mean.
- keymone 9y ago
  that's why it's called "relative", isn't it? Erlang and Clojure are undoubtedly a success for me in terms of productivity. Erlang is undoubtedly a success in a telekom industry since 90s, where extreme reliability is a basic requirement. isn't static typing promoted as a path to achieve reliability "once it's compiled"? go figure. imo, that majority of developers are being indoctrinated into false-OOP paradigm is rather unfortunate historical curiosity than a success indicator.
- iLemming 9y ago
  according StackOverflow survey - Clojure developers are the most well paid in the industry. What a miserable failures they are.
hellofunk 9y ago
Without a degree? I think that work experience in both dynamic and static languages is enough to offer a perspective. I don't have a degree in type systems but have done a lot of work in both Clojure and also C++ and Swift. I still can't decide if I prefer dynamic or static, they have nice tradeoffs.
whalesalad 9y ago
When you say degree are you specifically referring to an academic certificate? Honestly, real world experience and decades of software engineering trump anything you can get from a university.
- keymone 9y ago
  i agree, but not everybody does because i do find this sentiment repeated in such debates in different forms. for instance: https://i.warosu.org/data/g/img/0504/16/1442936748844.png https://i.warosu.org/data/g/img/0504/16/1442936748844.png
iLemming 9y ago
One thing that was a personal shocker for me to hear, from not just anyone, but Matthias Felleisen himself - "My research group has investigated the topic for almost 3 decades, and we came to the conclusion that Hindley-Millner doesn't really work". I was really baffled. That was at ClojureWest 2016. Here's the talk itself https://www.youtube.com/watch?v=XTl7Jn_kmio https://www.youtube.com/watch?v=XTl7Jn_kmio. That talk is not about Clojure - Felleisen doesn't even use Clojure. That happened before Clojure.spec was announced. I couldn't really figure out what exactly he was talking about. Not until later, when I tried Spec.

dgllghr 9y ago

Erlang & Elixir have a very interesting approach to typing. The support for pattern matching in function definitions provides a kind of structural typing for functions. And they both embrace typespecs for type analysis through a separate program analyzer. So a lot of the benefits of static typing are there without the static types.

keymone 9y ago
clojure has destructuring, which isn't quite pattern matching but it does help to detect wrongly shaped inputs. also dialyzer is amazing and i wonder if clojure.spec can be used like that.
- DigitalJack 9y ago
  clojure has pattern matching too. As a library. Much of the interesting functionality of clojure comes from libraries. (core.async, core.spec, core.match, etc..) The matching library is good. I wouldn't use it in a tight performance loop, but as a productivity booster it's very nice.

cutler 9y ago

Relative success? In what sense? As far as the jobs market is concerned Erlang/Elixir and Clojure have failed to make an impression. Ok, Wallmart, Bleacher Report etc. but search Indeed.com by job title and you'll get what I mean.

keymone 9y ago
that's why it's called "relative", isn't it? Erlang and Clojure are undoubtedly a success for me in terms of productivity. Erlang is undoubtedly a success in a telekom industry since 90s, where extreme reliability is a basic requirement. isn't static typing promoted as a path to achieve reliability "once it's compiled"? go figure. imo, that majority of developers are being indoctrinated into false-OOP paradigm is rather unfortunate historical curiosity than a success indicator.
iLemming 9y ago
according StackOverflow survey - Clojure developers are the most well paid in the industry. What a miserable failures they are.

hellofunk 9y ago

Without a degree? I think that work experience in both dynamic and static languages is enough to offer a perspective. I don't have a degree in type systems but have done a lot of work in both Clojure and also C++ and Swift. I still can't decide if I prefer dynamic or static, they have nice tradeoffs.

whalesalad 9y ago

When you say degree are you specifically referring to an academic certificate? Honestly, real world experience and decades of software engineering trump anything you can get from a university.

keymone 9y ago
i agree, but not everybody does because i do find this sentiment repeated in such debates in different forms. for instance: https://i.warosu.org/data/g/img/0504/16/1442936748844.png https://i.warosu.org/data/g/img/0504/16/1442936748844.png

iLemming 9y ago

One thing that was a personal shocker for me to hear, from not just anyone, but Matthias Felleisen himself - "My research group has investigated the topic for almost 3 decades, and we came to the conclusion that Hindley-Millner doesn't really work". I was really baffled. That was at ClojureWest 2016. Here's the talk itself https://www.youtube.com/watch?v=XTl7Jn_kmio https://www.youtube.com/watch?v=XTl7Jn_kmio. That talk is not about Clojure - Felleisen doesn't even use Clojure. That happened before Clojure.spec was announced. I couldn't really figure out what exactly he was talking about. Not until later, when I tried Spec.