I feel Elm with dependent types is probably just Idris. They seem to have
similar pragmatic philosophies, started from Haskell but have been
gradually breaking in ways to make things nicer, try to make good error
messages, etc. I'd love to see what TEA looks like in Idris though. And it
compiles
the first rule of Elm Type Club is you don't have to talk about types.
adding them back into the mix'd prob be hairy for a lot of people - then
again if you could do it in a v friendly way, there's lotsa benefits... I
heard Matz of Ruby fame talk about something similar once.
On Thu, 26 May 2016
I have thought about the idea of building a next gen elm with idris style
dependent types. To be called "Pine" of course ;)
Mind you I have neither the skills nor the time to do such a thing. But its
a cool idea
Zach
ᐧ
On Thu, May 26, 2016 at 9:56 AM, John Orford wrote:
A major issue is the "Maybe". In Java a type can be null so you always have
to check for that. WHile in Elm there is difference between "Maybe User"
and "User". So the type system will prevent any form of null pointer
exception.
Zach
ᐧ
On Tue, May 24, 2016 at 10:06 PM, Joey Eremondi
Are you interested in the actual type checking algorithms, or just the type
systems?
Big differences of the type systems:
* Elm has tagged union types, meaning that you can make a value that many
have one of many types, and pattern match on its possible variants.
* Elm has type inference, Java
Can someone /wittily/ sum up the experience of type checking in Java vs
something pure like Elm?
I feel purity, preciseness and descriptiveness is the main difference
somehow...
Java is too long in the distant past for me, but it's something I love
about Elm and never really cared for in