This looks like REBOL, a fascinating and impressive language that encourages the creation of DSLs (they call them dialects). It's a half-megabyte, zero-install, cross-platform executable that can run on Linux, Android, HaikuOS, Windows, Mac. From the homepage:
Most software systems have become needlessly complex. We rebel against that complexity, fighting it with the most powerful tool available, language itself.
Take a look at http://www.rebol.com/oneliners.html for some very impressive one-liners like the one below (downloads and prints a web page source):
At first glance it looks like a fun esoteric language, like Shakespeare [1]. Looking deeper, this language is more amazing then it leads on to be.
But for a little story time: I've started my first major project in Haskell while sick the past week, and it had me asking lots of questions:
Why do I have to add so many parentheses to get it to compile? Why can't I just pass monads into my functions? Why is every function not polymorphic? Why do I get so many pattern matching errors even with -Wall? :( The last one stings the most, believing my program would never crash. But every function being polymorphic really got me thinking, since instancing type classes with Monads was my first instinct. It felt so natural.
instance (Monad m, Num a) => Num (m a) where
(+) = liftM2 (+) -- ((return 123) + (return 321)) -- yipee!
...
I realize that type classes are global, and could have cascading problems. Even still, this made me want to prototype a language where all functions are polymorphic, but after looking over Avail, it seems to answer all my questions. It feels like a Lispy haskell; static type checking, with context sensitive polymorphism.
With this kind of polymorphism, one of my first thoughts was what the code compiles in an unintended way due to the way I specified. The metaphor I thought of to reason about this was: a copilot in the car tells you "Turn right here". Are they saying to turn right at the next intersection, or that this is the intersection that you should be turning at? This example shows succinctly shows that it's a problem with the language being define (i.e. the code), that "right" is overloaded to mean a direction, or immediately.
Anyways, I would love for anyone to weigh in, since I've been thinking about this a lot recently!
This is an interesting project, and I'm going to have to give a bit more time to it in order to wrap my head around a few things.
One thing that makes this a more difficult language for me to really get behind is the (required?) use of Unicode characters. That will take a bit of getting used to, as I am not in the habit of using ANY Unicode for anything.
I will still give this a chance, though. I'm always open to trying new ideas in the programming head-space.
I didn't find it yet, but I'd be very surprised if you couldn't just define a new "_is_" method/function/block/parse extension/whatever the hell it is to mean "::=". Or maybe "const_is_" if "_is_" wouldn't work.
Anyway, Avail looks very, very interesting, like it's a combination of Smalltalk, REBOL and Coq (although it's just an uneducated impression for now).
The goal appears to be the same. From the introduction:
"In the articulate programming paradigm, a programmer undertakes a project by developing a domain-appropriate lexicon and phraseology which are leveraged directly to implement a solution that is readily accessible to domain experts and software experts alike. Articulate programming is nothing more, and nothing less, than using the right language and conceptual framework for the job."
That's very similar to the common approach in Lisp of building the language up to reflect the problem domain.
Lisp, however, has the advantage of a minimal syntax that allows the programmer to extend the core language transparently. I'm not sure how possible that is in this language.
Smalltalk, Forth and Rebol are some other languages which allow for transparent language extension. There are others, too. Minimal syntax helps, as does homoiconicity, but Nimrod shows that you can do much even with fairly "mainstream" syntax.
Smalltalk, Forth and Lisp are mentioned specifically and in some detail in the Avail docs, see e.g. [0]. The designer apparently has had extensive experience with all of them. I can't find the reference now but at some point in the stuff I was reading last night was a mini-diatribe about Lisp, to the effect that it has "no syntax" and although it can indeed describe any problem, when you finish the solution doesn't "look like" the problem statement in any way.
Fun stuff. I imagine this must be a very difficult language to write a parser for. I think this makes a nice counterpart to the über-concise philosophies of Haskell and similar.
Looks like a lot of fun! I feel like a bit of a dick for asking this, but: isn't it really annoying needing to enter unicode-only characters (like →, ÷, “ and ”)?
It's unusual and I guess a lot of people actually would find it annoying, but I actually like it. The reason is there're many characters that are actually useful or even required if you intend to do typography "right", but not usually present on keyboard. I prefer "λ" to "lambda" (and I'm not greek), for instance. But I hate working with some abstractions over pure text, I want to see exactly what I typed, so I'm not big fan of all these vim plugins that show "λ" instead of "lambda" in Haskell, etc. God, I hate even tabs, because they look differently depending on their position in the line!
And typing unicode isn't annoying at all after you are used to it. I'm using XCompose all the time, so typing ½ or «» or λ is completely natural to me.
Most software systems have become needlessly complex. We rebel against that complexity, fighting it with the most powerful tool available, language itself.
Take a look at http://www.rebol.com/oneliners.html for some very impressive one-liners like the one below (downloads and prints a web page source):