Where are the new ideas in programming languages?

Question

I've recently been looking into the topic of programming languages and from what I've seen, few to none serious languages try making really "new" things that were not seen before their creation.

• Why do all more or less successful programming languages since 1980 or so just combine aspects of their predecessors?

I just can't believe that programming languages "can't get any better"..

Answer 1

If it aint broke, dont fix it.

Answer 2

Mankind hasn't invented the weel every two years again. Instead it is consistently improved. The same applies to programming languages.

Answer 3

Well, lisp has some stuff no one has copied yet and has been around forever (I'm looking at common lisp, specifically)

Answer 4

New languages are always coming, look at scala.

Answer 5

Computer Languages get new Ideas all the time but they still have to compile down to the same assembler instructions, so there can be no fundamental changes. In the end all your code will still consist of loops, memory segments (=> variables) and lots of jump instructions.

A new path of programming languages is Quantum Computing which might be a "new enough" idea for your taste.

Answer 6

Because we haven't yet figured out how to use many of the "old ideas" that have been around since at least 1970 really well, and there haven't really been any fundamentally new ideas that look like they'd completely change the game in a way that would make using them more useful than improving what we already have.

Answer 7

What about Subtext and Coherence?

Answer 8

Rather than being genuinely "new" I'd say that modern programming languages are gradually bringing in ideas which have been known about in computer science for decades, but haven't really hit industry - or which have previously been in niche languages, but not in mainstream ones.

For example, many of the ideas of LINQ aren't new per se - but for many developers LINQ will be their first experience of them.

Think of it as a process of "productising" previously "experimental" features. When I talked with Eric Lippert recently he said that the C# team was trying to avoid including "open research" ideas into the language - simply because they can't tell whether or not those ideas will pan out in the long run. There's plenty of room for experimentation in language, but probably not in mainstream languages.

Answer 9

Uh, Perl 6?

• active metadata on values, variables, subroutines, and types
• parameterised roles
• macros (implemented in Perl) that can rewrite AST or source code
• a grammar that is implemented in Perl 6 itself, and can be extended or modified by code
• multiple dispatch, distributive dispatch, delegation to the next item in the call chain
• grammars as first class objects in the language - can parse pretty much anything
• user-defined infix, prefix, circumfix operators that can use any Unicode character
• optional strong typing
• multimorphic equality tests
• hypothetical variables
• hyperoperators (SIMD processing of lists)
• coroutines
• lazy lists
• junctions (variables with multiple values)
• automatic parallelisation of code

Perl 6 takes the feature set of Perl, Ruby, Python etc., adds an optional strong type system, the OO of Smalltalk, and the functional abilities of Haskell - plus a few more innovations: just as Perl 5 brought regexes into the mainstream, Perl 6 makes non-regular grammars first-class citizens. And yes, it is under development; while not complete, it's moving fast: Rakudo Perl 6.

Answer 10

Programming languages can get much better! Have you looked at languages like Agda or Epigram? Of course these are research languages, so perhaps not what you would call 'serious languages'. However, if you are interested in where mainstream languages may go in the future they may offer some insight.