Slide 1
Slide 1 text
Deru a multilingual
programing language made with
python
Daniel Arango
Slide 2
Slide 2 text
Who am I?
DD
-I made this talking Colombia.
-PyCon Colombia Lightning Talk
-Novo Hamburgo University, Simultaneous
Multilingual Translation Systems
-I won 3 competitions at my uni with my
projects.
-Leader of ML group in Eafit University.
Slide 3
Slide 3 text
What is Deru
Lisp dialect in
De-> Deustche Spreche
(German language)
Ru->Русский язык
(Russian language)
My Motivations
Slide 4
Slide 4 text
John MCcarthy
Lisp Paper
LISt Procecing
What is lisp? Not a programming Language
Is a metaprogramming language
Slide 5
Slide 5 text
Lisp Dialects
Scheme Clojure Common
lisp
Emacs lisp
Slide 6
Slide 6 text
Examples
print("hello word")
(печать "привет
Мир")
print("hello word")
print("hello",input("how is your name"))
This is not an esoteric programing language, is more
powerful than that, but I didn't see the future for this
programming language, except for learning how to
make a programing language
Slide 7
Slide 7 text
More Examples of Deru
You have Control sctructures
Slide 8
Slide 8 text
More Examples of Deru
You can create Functions
Slide 9
Slide 9 text
More Examples of Deru
Mixing different natural languages
Slide 10
Slide 10 text
Notes
All is in parentheses. (Parentheses
mean it is a list).
Slide 11
Slide 11 text
Notes
The first element is the function.
The others are arguments.
(F a1 a2 a3 … ax)
F is a function name.
a1 ... ax are arguments.
Slide 12
Slide 12 text
Notes
Deru is not a pure Functional Programming
because it has side effects and monads are
not implemented
Is interpreted
Is not Complicated, sorry I mean
Compiled
Slide 13
Slide 13 text
https://github.com/kanaka/mal
How to make a lisp Dialect in python
Index
• step1_REPL
• step2_eval
• step3_lexical,syntaxis and
semantic analysis
• step4_Environment
• step5_Control_Structures
• … and more https://github.com/kanaka/mal
Slide 14
Slide 14 text
Repl (Read-Eval-Print Loop)
relp is the entry point to your programming
language
Slide 15
Slide 15 text
Eval
Python eval() is different from
this eval.
Lisp (F a1 a2 a3 … ax)
How to convert Sintaxis of the
functions
Python F ( a1 a2 a3 … ax )
The EVAL function processes Lisp code
as structured lists, treating them as
directly evaluable syntax. It interprets
constructs dynamically, supporting
macros, conditionals without an explicit
AST.
Slide 16
Slide 16 text
Tokenization
How look the code tokenized
Regular expression for tokenization
H
code NOT Tokenized
The process of separating the string into substrings (referred to as
tokens) organizes the information that is being passed to the parser.
Slide 17
Slide 17 text
Syntax's analysis
•The process for verifying that our
grammar is being followed
•Normally Syntax Analysis builds
an abstract syntax tree
•Here no, you traverse the tree
with the eval recursion
Slide 18
Slide 18 text
Grammar
• A grammar is a set of rules that describe how tokens can be combined to form valid
statements and expressions in the language.
Slide 19
Slide 19 text
S-expressions There are two kinds of S-expression, atoms and lists.
S-expressions express your program and manipulate it the same way an interpreter
would. This lets you add your own language features without modifying the compiler
or interpreter.
Slide 20
Slide 20 text
Semantics
Semantic Analysis makes sure that declarations and
statements of a program have a correct structure and
give the meaning to the tokens.
Slide 21
Slide 21 text
Type checking
We assign the types by validating the string with regular expressions,
regular expressions create Automata
Slide 22
Slide 22 text
Symbol table
In lisp dialects we don't use Symbol table
Slide 23
Slide 23 text
ENViroment
We save Quadrat function in the enviroment (python
dictionary)
Python Dictionary Complexity
O(1) get item
O(1) set item
O(1) delete item
We use dictonaries because their
operations have O(1) complexity
and let you change some
predefined functions programing
Slide 24
Slide 24 text
A programing language for all languages?
(2)add intermediate step for translation before access to the environment that
must be in English, in English because is and Interligua and most the
translators use it as intermediate language for optimizing translations
You have 2 ways to do this
(1) add all instructions translations to the environment
Slide 25
Slide 25 text
macros The rule or pattern that specifies how a certain
input should be mapped to a replacement
output.
Python doesn't have macros but we can create it with eval ()
Everything in lisp is a macro
Slide 26
Slide 26 text
Macros VS Functions
Slide 27
Slide 27 text
Variables and Functions
"Let" adds value to the symbol table
The third argument of the function is the code of the
function, which will insert the AST
We are inheriting the environment to manage public
and private variables.
Slide 28
Slide 28 text
Control structures
We use the "if" of deru passing the values to the "if" of
python , it is in the eval() Function
Slide 29
Slide 29 text
WHere are the "for or while"?
In functional programming
languages, 'while' and 'for'
are not used.
Slide 30
Slide 30 text
WHere are the "for or while"?
I know I'm a terrible
person for leaving `while`
commented out, but we
need to learn how to
make fast recursions.
Slide 31
Slide 31 text
Finally, we have a programming language
Slide 32
Slide 32 text
No content
Slide 33
Slide 33 text
No content
Slide 34
Slide 34 text
TCO (Tail call optimization)
Benefits:
• Avoids unnecessary stack
manipulation operations.
• No stack overflow
• Improved Performance
Slide 35
Slide 35 text
TCO (Tail call optimization)
Tail call optimization in Python is not natively
supported due to Python's recursion limit and the lack
of tail call optimization in its interpreter. However, it can
be simulated using a loop to replace recursive calls with
iterative ones and with externals tools.
Programing languages like haskell ,
Scala, Clojure, Erlang, and Elixir have TCO
Slide 36
Slide 36 text
"TCO (Tail call optimization)" simulated
focus on "while" addition
Slide 37
Slide 37 text
Trampolines
Trampolines allow to replicate tail call
optimization by wrapping function calls
into a special object using the
@trampoline decorator.
Slide 38
Slide 38 text
¿HOW to archive real TCO?
ctypes
Rust TCO cargo package didn't
work and had no documentation
With Ctypes you can connect C and
python using the TCO to C
Slide 39
Slide 39 text
Extra features
quotes
Try catch
Slide 40
Slide 40 text
Extra features
quotes
Quotes and comments
Slide 41
Slide 41 text
Common Problems when we create
a programing language
Slide 42
Slide 42 text
Finally, we have a programing language
Conclusions
Slide 43
Slide 43 text
end
jero98772
Follow me as
https://github.com/jero98772/Deru
References: https://pastebin.com/aynprhjU