IYH Dear Mariusz Postol

Programming Languages are like Swiss Army knives: versatile, capable of a wide range of tasks, and can solve almost any computational problem.

Non-Programming Languages (DSLs), like SQL, are specialized tools: extremely effective for specific tasks (e.g., database management) but not designed for broad, general-purpose computing.

A programming language is a set of instructions that a computer can execute directly to perform a wide range of tasks, from simple calculations to complex operations. These languages are designed to be broadly applicable across various domains and are typically so-called Turing complete, meaning they can theoretically compute almost any function. They also support control flow mechanisms like conditional statements (if/else) and loops (for, while), as well as variable assignment and manipulation, which enable complex logic and workflow management.

Non-programming languages, or DSLs, are designed for a narrow, specific purpose or domain. They are optimized for tasks within a particular field, such as database management, and are not intended for general computing tasks. DSLs often have a simplified syntax focused on the domain's needs, making them easier to learn for specialists in that domain. They may lack or severely limit features like conditional statements, loops, or variable manipulation, as these are not necessary for their specific application.

(Standard) SQL, or Structured Query Language, is an example of a DSL (basic HTML w/o CSS is another, JSON, YAML). Standard SQL is not designed to handle the broad spectrum of tasks that a general-purpose programming language can manage. It is specifically tailored for managing relational databases, allowing users to create, modify, and query database structures and data. While SQL can perform data manipulations and analyses, it is not as versatile as a general-purpose programming language. For instance, SQL's control flow elements, such as IF statements in stored procedures, are limited, variables in SQL serve a different purpose and are more limited in scope and functionality.

As a caveat, Extended SQL ie SQL extended with additional features, such as stored procedures and user-defined functions in database management systems like MySQL, PostgreSQL, and SQL Server, *is* Turing complete and would meet the definition of a programming language.

Daniyel Yaacov Bilar Thanks for the answer, but I am confused because from a separate discussion at https://www.researchgate.net/post/Is_SQL_domain-specific_language_DSL we can learn that SQL is a DSL. For me, it is also not clear the statement "A programming language is a set of instructions that a computer can execute directly". - what about types used in languages like Java, C++, C#, etc? Comments are welcome.

Daniyel Yaacov Bilar, again BTW, what is the difference between"Swiss Army knives:" and regular pocket knives? Is it just a brand, like Adidas(R), Nike(R), or Reebok(R) to name only selected ones?

IYH Standard SQL is a DSL. Extended SQL can be considered a bona-fide programming language.

The statement "A programming language is a set of instructions that a computer can execute directly" is an oversimplification, and you've correctly identified a key area where this definition falls short: typed languages with intermediate compilation or interpretation steps, such as Java, C++, C#, and many others. The 'directly' moniker was meant in contrast to non-programming languages in my post.

A more accurate but wordy non-math definition:

"A programming language is a set of instructions, along with its associated rules and semantics, that a computer can execute either directly, or indirectly through one or more stages of compilation, interpretation, or runtime environment processing, to produce almost any desired outcome that can be expressed by computable functions."

Everyday non-mathematical language is limited in expressing precisely what a Turing complete programming language is. These expressivity lacunae are, among other things, why the Bourbaki school emerged in mathematics in the 1930s, as an aside.

A classic Swiss Army knife has a dozen, sometimes two, sometimes more different blades or utensils. It is both a brand and a concept (like say Kleenex for tissue paper to blow your nose). There are of course non-Swiss Army knifes that are Swiss Army knifes in practice. I used this as a rough illustration for pre-lim orientation.

Computers are built to be able to execute some specific set of instructions. The machine level instructions are in binary. E.g. on one particular design, the "add" instructions might be 011010 (on most machines, 6 bits are used to indicate the instruction which allows for 64 different types of instructions). But nobody writes actual machine code; it is simply too hard to read and work with.

Instead people use a higher level language, a programming language, that is used to communicate the instructions we want executed. These languages are much easier for human to read and work with. Other programs translate the higher level language into something the machine can execute. An example statement in some programming languages could be

if (x > 0) y = 3;

"if" is a "keyword" that has a specific meaning. The following phrase inside the () must be something that evaluates to true or false. If the contents are variable x is greater than 0, then it evaluates to true, otherwise false. Nothing happens if it is false. If it is true, we execute the following statement "y = 3" which assigns the value of 3 to the variable y. The ; at the end is like a period in English, it indicates the end of the statement.

Compared to natural languages, programming languages have a more limited and very rigid syntax (syntax corresponds to a grammar in natural languages). The number of keywords in a programming language is much much smaller than the number of words in a natural language; natural languages have words for a huge number of concepts which have no use in a programming language. Natural languages are richer in other ways. Natural languages have metaphors, puns, and other things which programming languages lack.

Thanks for your contribution.

My concern is that today, even spoken languages may be recognized as programming languages thanks to generative AI. Hence, is there a condition we and our students can apply to recognize if we are using a programming language? The only thing that comes to my mind is the existence of an integrated development tool, which we can use to force a machine (computer) to behave as expected processing data or information (which one is correct data or information for you). Unfortunately, we have meta-languages and symbolic notations.

Glenn Rhoads you said: "The machine-level instructions are in binary." - do you mean a binary language?

A memory location in any computer can only be either "on" or "off", "on" amounting to be set to "1" and "off" amounting to being set to "0", that is, zero.

So all instructions in any program to be executed can only be stored in memory as series of 0s and 1s according to the conventions of binary coding. Only two values being available is what "binary" means.

André Michaud I agree "Only two values being available is what "binary" means" but it contradicts "memory as series of 0s and 1s.." I would appreciate any comments on this. Consider the CD as a memory and magnify the CD content (ask Google to "magnify the CD content"). Can you see a series of 0s and 1s.

Mariusz Postol

Dear Mariusz,

There is no contradiction. I will explain.

In 8-bit binary codes, for example, each instruction is assigned a block of 8 bits made of 0s and 1s. A program in 8-bit coding is stored as a series of such blocs of 8 bits in memory.

so, at execution time, the processor is instructed to start executing from the address where the beginning of the string of blocks is stored. It reads the series by jumps of 8 bits, and executes the programs as it goes until the end of the series of instruction is signalled.

If for example, one instruction that would mean "add" is represented by the series "00100110" and another instruction that would mean "subtract" is represented by "00100111" (arbitrary values here just for the example), the processor is hardwired to activate a different set of transistors, one set (activated by "00100110") that will execute an addition and is routed to another set of transistors (activated by "00100111") that will execute a subtraction.

All possibilities of 8-bit codes from "0000 0001" to "1111 1111" are available. in 16 bit codes, then the number of locations is doubled, and so on.

A CD recording for a movie picture or music piece is something different, a continuous series meant to reproduce digitally continuous analog wave signals

André Michaud Unfortunately, your explanation doesn't help. I want to engage you to say something like: "Any two values being available is what "binary" means". In other words, the binary representation must be founded on a set containing only and only two signs. It could be {0,1}, pits and lands, voltage level, current level, magnetic residue, etc. It makes it very easy to change the representation and read the CD into the memory converting pits and lends to voltage level, providing you know an agreement on how to convert pits and lands into a voltage level.

I believe that the discussion about the binary format is important but far beyond the main thread topic. Let me recall: "What is the difference between programming and non-programming languages?" Let me give you an example: Is XML a programming language? To answer you must recall, for example, XAML. I would appreciate reference to any existing academic publications.

Mariusz Postol

Of course, 0s and 1s in computer treatment are related to voltage.

In 5 volts TTL circuits, memory chips record 5 volts at locations that represent 1 and zero volt at locations that represent 0.

In 3.5 Volts circuitry then 1s are represented by a location set at 3.5 volts and 0s by zero volt at this location.

With respect to the main topic "What is the difference between programming and non-programming languages?"

So-called "non-programming" languages are only sets of specialized macros dedicated to specialized tasks. They are fundamentally also "programming languages", but dedicated programming that cannot process data outside the special restricted range of processes that they have been designed to process, as explained by other contributors.

General programming languages such as Fortran, Forth, C++, python, etc... provide the complete lower level instructions sets, all reducible to the specific assembly language of each processor type, each instruction of which assembly language ultimately resolving to the binary code assigned to each possible mathematical or logical operation, that allow programming any application that linear processing computers can address.

Greetings

Programming languages are used to write instructions that a computer can understand to perform specific tasks, such as Python and Java.

Non-programming languages are natural languages used for communication between humans, such as Arabic and English, and they are not used to directly instruct computers.

Hadeer Majed, do you think XML may be classified as a non-programming language according to your definition?

Mariusz Postol

XML can be considered a non-programming language because it is used to store and organize data, rather than to execute instructions or operations like programming languages.

Best regards

Helmi Hamdi, thank you for your contribution. My concern is that XAML is compliant with XML and is used to describe functionality. This proves that XML may be used to describe not only data but also functionality. What about SQL - is it a programming language?