Hacker News Reader

You're right. Its much easier to have domain specific languages specially built for certain problems. The appeal of a single language, I guess, is that you only need to learn one language and you can work on all problems. Then again, if you have good support for macros and other such "niceties" then you can remold the language to be domain specific anyway, so I think having a single super language isn't so far fetched and is actually not a bad idea really.

View on HN

I can't help but think of Erlang. Erlang is by no means a DSL, but it is very well suited (both in terms of language features and its Virtual Machine) to writing distributed, concurrent applications. Why manipulate a hyopthetical language so that it is functional, has message-passing, allows hot-swapping of code, enforces single assignment etc. when a language with all those features already exists? And once you have manipulated a language to that extent haven't you created another language anyway, eliminating the possible benefits of having a single, malleable language?

Well, for this hypothetical scenario, its for the ability to write all your code within the one language, but tuning each code block or abstraction to the particular problem being solved at that precise time.

In real life, you're right. It doesn't make sense to try and mash every language together just because. Its also potentially not possible anyway, since theres too many conflicting design decisions - though the ability to mold a language, through powerful abstractions and macros, like Lisp does, does indeed solve a lot of this.

Reply on news.ycombinator.com