I want to make a programming language...

nickeljorn

I want to make a programming language, but I don't know how.

"Just what is the Summer Solstice, besides the day I melt into a puddle of fuzzy puppy fluff?”-Isabelle, Animal Crossing New Leaf

NitroCipher

A programming language of your own? Or an interpreter of an existing one?

I hope my post helped you in some way! Post count: 500+

Current project: [s3Blocks: scratchblocks rewritten for Scratch 3.0] ::#4b4a60 //https://scratch.mit.edu/discuss/topic/290031/ Basically done!

This is my signature identifier “aWFtbml0cm9jaXBoZXI=”

nickeljorn

NitroCipher wrote:
A programming language of your own? Or an interpreter of an existing one?

A new programming language.

"Just what is the Summer Solstice, besides the day I melt into a puddle of fuzzy puppy fluff?”-Isabelle, Animal Crossing New Leaf

MegaApuTurkUltra

Racket

$(".box-head")[0].textContent = "committing AT crimes since $whenever"

TheAspiringHacker

Do you want to make a specification (a formal document outlining the definition of the language) or an implementation (software that understands programs written in the language)?

Long live Kyoto Animation!

ScratchMan544

If you're actually looking at the implementation of programming languages:

I'll start with an example from a fake language. The program string is this:

let i = 0;
while i < 100 {
    print("Number " + i.to_string());
    i += 1;
}

The first step is to convert the program into small lexical units, called tokens. The internal representation for tokens might look something like this:

KEYWORD_LET
IDENTIFIER "i"
ASSIGN
NUMERIC_LITERAL 0
SEMICOLON
KEYWORD_WHILE
IDENTIFIER "i"
LESS_THAN
NUMERIC_LITERAL 100
OPEN_BRACE
IDENTIFIER "print"
OPEN_PARENTHESIS
STRING_LITERAL "Number "
PLUS
IDENTIFIER "i"
ATTRIBUTE "to_string"
OPEN_PARENTHESIS
CLOSE_PARENTHESIS
CLOSE_PARENTHESIS
SEMICOLON
IDENTIFIER "i"
PLUS_EQUALS
NUMERIC_LITERAL 1
SEMICOLON
CLOSE_BRACE

Next, you convert this to a tree that represents the structure of the program. This requires knowing about things like operator precedence.

[
    (declare "i" (numeric-literal 0)),
    (while-loop (less-than (var "i") (numeric-literal 100) [
        (call-function (var "print") [
            (plus (string-literal "Number ")
                  (call-function (property "to_string" (var "i")) []))
        ]),
        (plus-assign "i" 1)
    ]
]

Now, at this point you have a few options:

You can execute this tree directly, which is oftentimes slower but simpler and more direct (this is an interpeter).
You can translate it into an even lower-level representation (this is a compiler). This often requires going through at least one more intermediate representation.

Last edited by ScratchMan544 (Dec. 8, 2017 08:09:52)

_=(lambda _:lambda __:_(__))(lambda _:getattr(_,(
    lambda _:_[:2]+str(print.__call__)[0b10011:(1+1<<1+1+1)+(1<<1+1)+(1<<1)+1]+_[-2:]
)(__name__)))(eval)
(lambda _:lambda __:_(__))(lambda _:_(_(
    __import__(dir(__builtins__)[((1<<1+1)<<1+1+1)+(1+1<<1+1+1)+(1+1<<1)+(1<<1)][:3].lower()),
    print.__doc__[46:52]),open(__file__).write.__str__()[17:22]))(_("getattr"))((
    lambda _:lambda __:_(_,__))(lambda _,__:""if __==0else chr(__%128)+_(_,__//128))(963149002634454890336513358634316810781103160855182366005237514)[::-1]
)

scorebat

Why would you want to create a new language?
There are already so many of them.
Scorebat

Scorebat

herohamp

scorebat wrote:
Why would you want to create a new language?
There are already so many of them.
Scorebat

For education. Depending on how you write your language, you may need to write a system which compiles it to either binary, or another language that compiles to binary. Or you would need to write a parser

My website hamptonmoore.com
UMass Amherst All Campus Makerspace
Supporting the wife's SEO
Click for a Scratch based programming language I made

_nix

ScratchMan544 wrote:
Next, you convert this to a tree that represents the structure of the program. This requires knowing about things like operator precedence.

It actually doesn't. If you want your programming language to have automatic order of operations (e.g. knowing that 4 + 5 * 2 is 14 and not 18), you'll have to figure out how to implement that (which is operator precedence), but it's not inherently necessary when you're making a programming language!

══ trans autistic lesbian enbydoggirls // 16 17 18 19 20, she/they ════
sparrows one word to the paragraph // <3 // ~(quasar) nebula

NickyNouse

_nix wrote:
ScratchMan544 wrote:
Next, you convert this to a tree that represents the structure of the program. This requires knowing about things like operator precedence.
It actually doesn't. If you want your programming language to have automatic order of operations (e.g. knowing that 4 + 5 * 2 is 14 and not 18), you'll have to figure out how to implement that (which is operator precedence), but it's not inherently necessary when you're making a programming language!

my order of operations is ltr thank you very much

MegaApuTurkUltra

_nix wrote:
ScratchMan544 wrote:
Next, you convert this to a tree that represents the structure of the program. This requires knowing about things like operator precedence.
It actually doesn't. If you want your programming language to have automatic order of operations (e.g. knowing that 4 + 5 * 2 is 14 and not 18), you'll have to figure out how to implement that (which is operator precedence), but it's not inherently necessary when you're making a programming language!

muh lisp

(+ 4 (* 5 2))
(* (+ 4 5) 2)

infix notation was a mistake and we all know it

$(".box-head")[0].textContent = "committing AT crimes since $whenever"

bobbybee

MegaApuTurkUltra wrote:
_nix wrote:
ScratchMan544 wrote:
Next, you convert this to a tree that represents the structure of the program. This requires knowing about things like operator precedence.
It actually doesn't. If you want your programming language to have automatic order of operations (e.g. knowing that 4 + 5 * 2 is 14 and not 18), you'll have to figure out how to implement that (which is operator precedence), but it's not inherently necessary when you're making a programming language!
muh lisp
(+ 4 (* 5 2))
(* (+ 4 5) 2)
infix notation was a mistake and we all know it

Hmmmmmph. I like Scheme. Brian broke me :^)

“Ooo, can I call you Señorita Bee?” ~Chibi-Matoran

TheUltimatum

bobbybee wrote:
MegaApuTurkUltra wrote:
_nix wrote:
ScratchMan544 wrote:
Next, you convert this to a tree that represents the structure of the program. This requires knowing about things like operator precedence.
It actually doesn't. If you want your programming language to have automatic order of operations (e.g. knowing that 4 + 5 * 2 is 14 and not 18), you'll have to figure out how to implement that (which is operator precedence), but it's not inherently necessary when you're making a programming language!
muh lisp
(+ 4 (* 5 2))
(* (+ 4 5) 2)
infix notation was a mistake and we all know it
Hmmmmmph. I like Scheme. Brian broke me :^)

Racket.

solve the cube

bobbybee

TheUltimatum wrote:
bobbybee wrote:
MegaApuTurkUltra wrote:
_nix wrote:
ScratchMan544 wrote:
Next, you convert this to a tree that represents the structure of the program. This requires knowing about things like operator precedence.
It actually doesn't. If you want your programming language to have automatic order of operations (e.g. knowing that 4 + 5 * 2 is 14 and not 18), you'll have to figure out how to implement that (which is operator precedence), but it's not inherently necessary when you're making a programming language!
muh lisp
(+ 4 (* 5 2))
(* (+ 4 5) 2)
infix notation was a mistake and we all know it
Hmmmmmph. I like Scheme. Brian broke me :^)
Racket.

Well sure, that's my preferred Scheme of choice, but hey

“Ooo, can I call you Señorita Bee?” ~Chibi-Matoran

MegaApuTurkUltra

bobbybee wrote:
TheUltimatum wrote:
bobbybee wrote:
MegaApuTurkUltra wrote:
_nix wrote:
ScratchMan544 wrote:
Next, you convert this to a tree that represents the structure of the program. This requires knowing about things like operator precedence.
It actually doesn't. If you want your programming language to have automatic order of operations (e.g. knowing that 4 + 5 * 2 is 14 and not 18), you'll have to figure out how to implement that (which is operator precedence), but it's not inherently necessary when you're making a programming language!
muh lisp
(+ 4 (* 5 2))
(* (+ 4 5) 2)
infix notation was a mistake and we all know it
Hmmmmmph. I like Scheme. Brian broke me :^)
Racket.
Well sure, that's my preferred Scheme of choice, but hey

me too thanks

$(".box-head")[0].textContent = "committing AT crimes since $whenever"

TheUltimatum

To add to the topic.. I'd recommend this tutorial by Jack Crenshaw. I think it's a really great intro to compiler design. Afterwards if you liked his tutorial I'd reccommend getting Compilers: Principles, Techniques and Tools I own the second edition bought it on amazon for 2$ used not bad at all and the book is great reading material. Though you may have a fun time porting the pre standard C to modern C. A very a good read I'm about a quarter of a way through and already learning all kinds of interesting things.

solve the cube

Discuss Scratch