Are all languages basically the same?

Question

Recently, i had to understand the design of a small program written in a language i had no idea about (ABAP, if you must know). I could figure it out without too much difficulty.

I realize that mastering a new language is a completely different ball game, but purely understanding the intent of code (specifically production standard code, which is not necessarily complex) in any language is straight forward, if you already know a couple of languages (preferably one procedural/OO and one functional).

Is this generally true? Are all programming languages made up of similar constructs like loops, conditional statements and message passing between functions? Are there non-esoteric languages that a typical Java/Ruby/Haskell programmer would not be able to make sense of? Do all languages have a common origin?

Answer 1

I'll direct you to Beating the Averages by Paul Graham. You may or may not want to read the whole thing, but for the relevant part search for the heading "The Blub Paradox". Mr. Graham can't be bothered to put anchors in his wall of text, so I can't directly link to it.

Answer 2

I would say that a language encodes meaning. If the meaning has any sense in some context then all languages that could express the meaning could be said to be equivalent limited by the meaning and the context.

If you limit that context to a standard von newman (I think) machine then the computational meanings of changing memory and computing in a cpu could be said to be the origin - and possibly the only meaning that all the languages have in common. All other things are abstraction built on these.

Answer 3

No, some languages are more powerful than others:

• More powerful: Smalltalk, Lisp, Haskell, Icon, Perl

• Less powerful: Simula-67, APL, ML, Snobol, Bourne shell

Mastering the intent of an Icon or Prolog program is nearly impossible until you understand each language's nonstandard model of evaluation.

Answer 4

Hardware description languages are programming languages, but they are conceptually very different. Try VHDL or Verilog on for size. They are common for programming FPGAs. (Ok, so they're not processors, but they are general purpose computing devices. And such should be considered valid hardware for computer science topics.) You have to explicitely make things occur serially. It's a completely different model. You have think of things occuring in parallel as the rule not the exception. For loops in verilog expand into parallel hardware. So the "expected" behavior might not be what you expect.

Answer 5

The basics of most procedural languages are pretty much the same.

They offer:

• Scalar data types: usually boolean, integers, floats and characters
• Compound data types: arrays (strings are special case) and structures
• Basic code constructs: arithmetic over scalars, array/structure access, assignments
• Simple control structures: if-then, if-then-else, while, for loops
• Packages of code blocks: functions, procedures with parameters
• Scopes: areas in which identifiers have specific meanings

If you understand this, you have a good grasp of 90% of the languages on the planet. What makes these languages slightly more difficult to understand is the incredible variety of odd syntax that people use to say the same basic things. Some use terse notation involving odd punctuation (APL being an extreme). Some use lots of keywords (COBOL being an excellent representative). That doesn't change much.

More interesting procedural languages offer

• Nested or lexical scopes, namespaces
• Pointers allowing one entity to refer to another, with dynamic storage allocation
• Packaging of related code: packages, objects with methods, traits
• More sophisticated control: recursion, continuations, closures
• Specialized operators: string and array operations, math functions

The languages which are harder to understand are the non-procedural ones:

• Purely functional languages, with no assignments or side effects
• Logic languages, such as Prolog, in which symbolic computation and unification occur
• Pattern matching languages, in which you specify shapes that are matched to the problem, and often actions are triggered by a match
• Constraint languages, which let you specify relations and automatically solve equations
• Hardware description languages, in which everything executes in parallel
• Domain-specific languages, such as SQL, Colored Petri Nets, etc.

There are two major representational styles for languages:

• Text based, in which identifiers name entities and information flows are encoded implicitly in formulas that uses the identifiers to name the entities (Java, APL, ...)
• Graphical, in which entities are drawn as nodes, and relations between entities are drawn as explicit arcs between those nodes (UML, Simulink, LabView)

The graphical languages often allow textual sublanguages as annotations in nodes and on arcs. Odder graphical languages recursively graphs (with text :) in nodes and on arcs. Really odd graphical languages allow annotation graphs to point to graphs being annotated.

Most of these languages are based on a very small number of models of computation:

• The lambda calculus (basis for Lisp and all functional languages)
• Post systems (or string/tree/graph rewriting techniques)
• Turing machines (state modification and selection of new memory cells)

Given the focus by most of industry on procedural languages and complex control structures, you are well served if you learn one of the more interesting languages in this category well.

I highly recommend learning Scheme, in particular from a really wonderful book: Structure and Interpretation of Computer Programs. This describes all these basic concepts. If you know this stuff, other languages will seem pretty straightforward except for goofy syntax.

Answer 6

It all depends. If you define the language as just syntax, then they are fairly similar. (Excluding the Brainf*** language)

But people do favour languages over another because of slight differences such as how verbose a language is, what frameworks the language supports, whether the language is extensible, whether there is a big community behind it, etc. These small differences tend to make a big when you use the same constructs over and over again in your code.

Also, if there is a common origin, it would probably be C.

Answer 7

Depends on what you mean by "basically." All languages of any flexibility are Turing-complete. In that sense: yes, they are all basically the same.

On a low level, they all perform similar sequences of operations and all Windows, Linux, and (recent) OS X stuff all runs on Intel-compatible processors using the same instruction sets. In that way they are also basically the same.

I realize that you sorta defined "basically" in your question, but to really answer it, that definition will have to be much more refined. In many ways they are all alike. In many ways they are distinct. It's all too easy to say "it depends." If you take either extreme, the question will likely not answer what you intend it to so where that line is drawn is crucial to answering your question as you intend it.

Answer 8

Mature languages generally have a few goals, and they make tradeoffs where they sacrifice one thing for another. A general-purpose language can be used for anything, but no language can ever excel in every area. A few examples:

C attempts to be an ideal systems programming language. To this end it sacrifices readability and safety for low-level control and speed.

Python aims to be an ideal scripting language. To this end it sacrifices speed and verifiability for productivity and portability.

Haskell attempts to be a safe, mathematically pure language. To this end it sacrifices learnability and convention for verifiability and reliability.

These sacrifices and benefits make a huge difference in the language. Yes, most programming languages can be used for anything that can be done by a computer, but none of those same languages should be used for everything. All of the languages above are ones I would choose for certain tasks but not for others. If I were programming an operating system I would choose C. If I were writing a backend for a website, I would use Python. And if I were writing a financial system I would use Haskell.

In the end, your choice as a programmer is what the right tool for the job is.

Answer 9

Piet is pretty different. :)

Answer 10

I would not consider things like Lisp and Prolog to be anything like ordinary languages.