arkscript-lang / ark Goto Github PK

Is your feature request related to a problem? Please describe.

I followed the build steps on macOS, until cmake --build build. It ends with:

~/lisp/Ark/include/Ark/Utils.hpp:10:10: fatal error: 'filesystem' file not found
#include <filesystem>
         ^~~~~~~~~~~~

The comment on the next step ("works on Linux and on Windows") implies that the project is not meant to be built on macOS.

Is that the case, or perhaps I'm missing a dependency?

Describe the solution you'd like

I'd love to be able to play with ArkScript on macOS.

Additional context

As context, I'm pretty much a beginner with C++, so I have no clue whether filesystem is a platform-specific thing.

Instead of having to send all the argument using the command line (which can lead to a pretty long command), we should be able to have a project.arkopt for example, with a list of options which should be triggered.

For example:

cool-project/
|___ cool-project.arkopt
|___ thing.ark
|___ ...

cool-project.arkopt

# file to run
thing.ark
# options
-fno-function-arity-check
-f...

And when we run Ark cool-project it will search for a cool-project ArkScript file, but not find it, thus it should look for a cool-project.arkopt and use its content as the options of the command line.

eg. the VM should display the name of the variable which is redefined by let

In this moment, I'm working on msgpack module for Ark, and I need an clear and explicit way for define custom types for modules (plugin).

m_environment.reserve(64); should be m_environment.reserve(length); in rework/src/VM/Frame.cpp

Instead of using shared pointer, we should have raw pointers to elements in a std::list owned by the VM. (see comment)

Description related data owned by the VM should also be moved to a lighter structure (Ark::State)

• remove all the empty destructors
• do not create nil/false/true as they are available in Ark::internal::FFI

eg. the parser and compiler should detect errors like (fun output (@ mylist index)) which isn't a valid function definition

At this point where modules are starting to be more numerous, we should (imo) consider having an ArkScript package manager.

Won't be for the 3.1.0, more like a 3.2.0 or even a 3.x.0

Should we put the modules in another repository?

I'm asking this because the repository will definitely grow a lot with all the modules we are going to add, and when someone wants to start experimenting on a project, often they don't like having to download a bunch of things that they aren't going to use.

Having like a module store would be quite cool, what do you think?

Describe the bug
When launching Ark tests/games/snake/snake.ark -c, it crashes because it can't create a token from (() at line 13, char 38 (this token isn't in the file).

To Reproduce
Steps to reproduce the behavior:

1. Clone the repository
2. git checkout rework
3. cmake -H. -Bbuild -DCMAKE_BUILD_TYPE=Release && cmake --build build
4. build/Ark tests/games/snake/snake.ark -c

Expected behavior
No errors expected.

Desktop (please complete the following information):

• OS: Linux Mint 18 64bits

Additional context
Must be related to the new lexer, the parser is just reporting a bug because it uses the lexer

If we move description related data to an Ark::State as suggested in #35, we should be able to share a single state between multiple virtual machines, thus allowing multi threading

Describe the bug
This test is failing, and so is switch.ark, since it's using the same method for defered calls.

{
    (let egg (fun (bar) (print bar)))
    (let data ["Iron Man" "is" "Tony Stark"])
    (let callbacks [])

    (let add-callback (fun (d)
        (fun () (egg d))
    ))

    (let acc 0)
    (while (!= acc (len data)) {
        (let d (@ data acc))
        (set callbacks (append callbacks (add-callback d)))
        (set acc (+ 1 acc))
    })

    (set acc 0)
    (while (!= acc (len callbacks)) {
        ((@ callbacks acc))
        (set acc (+ 1 acc))
    })
}

To Reproduce
Steps to reproduce the behavior:

1. download the latest version, build it
2. compile tests/callbacks.ark
3. execute it
4. See error

Expected behavior
We should see

Iron Man
is
Tony Stark

in the shell, not an error.

Desktop (please complete the following information):

• OS: Linux Mint 18 x64

Additional context
Must be related to the parser

The program {(while (= 1 1) "foo" "bar")} segfaults the compiler.

(gdb) run foo.ark 
Starting program: /tmp/Ark/build/Ark foo.ark

Program received signal SIGSEGV, Segmentation fault.
0x00007ffff7bce3c0 in __memcmp_avx2_movbe () from /usr/lib/libc.so.6
(gdb) bt
#0  0x00007ffff7bce3c0 in __memcmp_avx2_movbe () from /usr/lib/libc.so.6
#1  0x00007ffff7eda920 in std::char_traits<char>::compare (__n=<optimized out>, __s2=0x555555624a70 ")", 
    __s1=<optimized out>) at /build/gcc/src/gcc-build/x86_64-pc-linux-gnu/libstdc++-v3/include/bits/char_traits.h:325
#2  std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >::compare (this=0x7fffffffc6a8, 
    __s=0x555555624a70 ")")
    at /build/gcc/src/gcc-build/x86_64-pc-linux-gnu/libstdc++-v3/include/bits/basic_string.tcc:1425
#3  0x0000555555579e4b in bool std::operator==<char, std::char_traits<char>, std::allocator<char> >(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, char const*) ()
#4  0x0000555555579fe5 in bool std::operator!=<char, std::char_traits<char>, std::allocator<char> >(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, char const*) ()
#5  0x000055555560171b in Ark::Parser::parse(std::__cxx11::list<Ark::internal::Token, std::allocator<Ark::internal::Token> >&, bool, bool) ()
#6  0x0000555555600156 in Ark::Parser::feed(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&) ()
#7  0x00005555555ea93c in Ark::Compiler::feed(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&) ()
#8  0x000055555555e67b in Ark::compile(bool, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&) ()
#9  0x000055555557c1da in Ark::VM_t<false>::doFile(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&) ()
#10 0x000055555555f711 in main ()

Additionally, I think that the benchmark files are missing from this repository and I had to modify build.sh because it assumed my system runs GCC 8, but Arch provides GCC 9, and ((f f) x) isn't recognised as valid syntax by the reader, but the segfault is my biggest concern now.

Is your feature request related to a problem? Please describe.
We should be able to choose what we need to "import" from another scope.

Describe the solution you'd like

(fun (name age)
    (fun (foo &name) ())
)

Arguments starting with & aren't arguments but captured values.

Describe the bug
The REPL segfaults when:

1. The input is only spaces
2. The input is only a comment (starts by optional spaces and then #)

Expected behavior
Same behavior as if the user had directly pressed <ENTER>: "nothing" (new line and new prompt)

Relevant GIF

Notes (1 & 2):
Here is the code handling the REPL when the user simply presses <ENTER>:

if (line.length() == 0)
{
    std::cout << (new_command ? new_prompt : continuing_prompt);
    continue;
}

It only checks that the line is empty.
Maybe it should instead strip whitespaces ahead (spaces and tabs notably), and then conclude:
If the resulting line starts has a length of 0 or starts with # then: ...

I'm only doing an issue as I have no idea how to do this in C++

EDIT:
The GIF also shows how {} or the equivalent (begin ) block also causes the REPL to segfaults. But this is the current behavior of Arkscript itself and isn't "the fault" of the REPL implementation.

It should return a reference to an object in the global scope, setting nil as the value if it doesn't exist yet

Is your feature request related to a problem? Please describe.
We should be able to use Ark file to automatically compile the file only if needed, and then execute it on the virtual machine.

Describe the solution you'd like
Check if there is a bytecode file in the Ark cache for the given input file, and if it is older than the given file, recompile. Otherwise (if not older), launch the bytecode file located in the Ark cache (maybe provide something like Python, a __arkcache__ folder). If the file isn't in the Ark cache (or if the Ark cache doesn't exist yet), compile it, put it in the Ark cache, and execute it.

This issue will be used to keep track of the needed builtins and std lib functionnalities for the 3.1.0

List

• (zip list1 list2)
• (map func list1)
• sorting algorithms
• reverse array
• find object in array
• delete element at index and return new list
• (slice array from_index to_index step)
• implement forEach (already done in examples/for-each.ark, but it can be greatly improved)

Dictionnary

• dictionnary implementation using closures see the json module

String

• regexes
• reverse string
• find substring in string
• (slice string from_index to_index step)
• (removeAtStr string index)

Network

• create socket
  • TCP/IP
  • UDP
• close socket
• send data
• receive data
• make ARP request
• make ICMP packets

Mathematics

• trigonometrics
  • cos, arccos
  • sin, arcsin
  • tan, arctan
• exp, log
• pow (can be implemented in ArkScript with exp and log)
• sqrt (can be implemented in ArkScript with pow)
• ceil, floor, round
• sum (elements from list)
• is infinite, is nan
• pi, e, tau, inf, nan

SFML

• Graphics
• Audio
• Window
• System
• Network

Not part of the standard library anymore, was moved to https://github.com/SuperFola/arksfml

It erases frames but shouldn't

Is your feature request related to a problem? Please describe.
As i already said, i'm doing a mini Os in Ark (#28) so it also could be useful to have an CLS method !

Describe the solution you'd like
CLS Method

Describe alternatives you've considered
(clearScreen)
(clearScreen 5) A specific number of lines
Additional context
Mmmh, For a Lot of console thing like as a Command line OS ?

Is your feature request related to a problem? Please describe.
The representation (through print) of lists are their content in parentheses.
For example (print [1 2]) outputs ( 1 2 )

This representation looks like a block (of code) and doesn't really hint that the object considered is a list.

Describe the solution you'd like
I think it would be better if the representation matched the syntactic sugar used in
the programs: [a b c]

Describe alternatives you've considered
[ a b c ], or (list a b c) (according to what the syntactic sugar refers to).

Ark-windows-x64.exe Bug
After moving it to C:\Program Files (x86)\Ark\bin , i can't use it to launch the .ark file
that's kinda weird, because it worked before (when the file was on my Desktop)

It's pretty messy right now

Could be based upon cpp httplib

Is your feature request related to a problem? Please describe.
We must be able to read public (immutable) data from a closure.

Describe the solution you'd like
eg: (print closure.field)

• standard library
• builtins

ArkDoc: https://github.com/ArkScript-lang/ArkDoc

Describe the bug
The parser is parsing { (let a nil) (let b false) (let c true) } as ( Begin ( Let a nil ) Begin ( Let b false ) Begin ( Let c true ) ). It's not a problem with the syntactic sugar, since we have this:

[  Info ] (Parser) After applying sugar:
   1 | (
     | begin
   2 | (
     | let
     | a
     | nil
     | )
   3 | (
     | let
     | b
     | false
     | )
   4 | (
     | let
     | c
     | true
     | )
   5 | )

To Reproduce
Steps to reproduce the behavior:

1. compile Ark
2. Launch it on nearly any test file with at least 2 blocks in a begin node

Expected behavior
The parser shouldn't put that many begins (one before each block, WHY?).

Desktop (please complete the following information):

• OS: Linux Mint 18 x64

should also mark the Value as const

https://github.com/SuperFola/Ark/blob/rework/include/Ark/Compiler/Compiler.hpp#L38

• rshift
• lshift
• binary or
• binary and
• binary not ~
• not

Since the language is evolving, its documentation should evolve as well. The language itself needs to be documented, but the standard library and the Virtual Machine/Bytecode as well

Describe the bug
toUpperCase and toLowerCase returning "Unknown Error"

To Reproduce
Steps to reproduce the behavior:
1.Import Case.ark
2. try (toLowerCase "Some StringYY sTuffY") or another string NB : you can also do this with toUpperCase
3. Enjoy the bug
Note : sometimes it returns another error like bad allocation
Expected behavior
Returning the string to lowercase (or to upper)

Screenshots
If applicable, add screenshots to help explain your problem.

Desktop (please complete the following information):

• OS: Windows 10 64bits
• Version 3.0.7

Additional context
Seems to happen with the latest version of Case.ark (i tried with an old version and it works)

Describe the bug
When sending End of File to the REPL, it prints indefinitely the prompt >>

To Reproduce
Steps to reproduce the behavior:

1. Ark -r
2. Ctrl + D (^D)
3. >> is printed indefinitely

Expected behavior
The REPL exits

Desktop (please complete the following information):
OS: Arch Linux
Ark compiled with g++ (GCC) 9.2.0 from branch dev

Describe the bug
i Tryed to run unittest.ark, and it says "ParseError: Couldn't find file Switch.ark"
Note: i run it without the --lib argument
To Reproduce
Steps to reproduce the behavior:

1. Go to tests directory
2. Launch unittest.ark
3. See error

Expected behavior
I Expect to import the module, wich is already installed with Ark Releases

Screenshots

Desktop (please complete the following information):

• OS: Windows
• Version : 3.0.5
  Additional context
  I think it's the import bugging because it happens on the first import

I think we should consider adding macros to the language, since it would have been (imo) very useful for some programs, and also to extend the language (eg if we want to define a forEach "keyword", which wouldn't use a function behind but directly copy the code doing the forEach thing)

Should be able to type in code, wait for count("(") == count(")") before compiling.
Store code in buffer, compile and execute after each line.

Is your feature request related to a problem? Please describe.
Well, i'm trying to do a sort of Command line OS (fo no reason), and i think it could be nice if i can list all files of the current Directory/Path

Describe the solution you'd like
Add a method to list all the files in a array

Describe alternatives you've considered
(listAllFiles ".") Files of the current directory (also default argument)
(listAllFiles "C:\Users\Natend\Desktop\Programmation") Mmh maybe not \
Additional context
For creation of system that can see files (a.k.a : OS, IDE, Lots of things)

We need to have at least:

• write file
• read file
• file exists?
• string concatenations
• type
• hasfield

Rules list is quite simple:

string: ^"[^"]*
number: ^((\+|-)?[[:digit:]]+)(\.(([[:digit:]]+)?))?
operator: ^(\+|\-|\*|/|<=|>=|!=|<|>|@=|@|=|\^)
identifier: ^[a-zA-Z_\u0080-\uDB7F][a-zA-Z0-9_\u0080-\uDB7F\-?']*
capture: ^&[a-zA-Z_\u0080-\uDB7F][a-zA-Z0-9_\u0080-\uDB7F\-?']*
getfield: ^\.[a-zA-Z_\u0080-\uDB7F][a-zA-Z0-9_\u0080-\uDB7F\-?']*
skip: ^[\s]+
comment: ^#.*
shorthand: ^[']
mismatch: ^.

To be able to create packed messages to send on the internet when the network module will be available.

Is your feature request related to a problem? Please describe.
Well nope, but it could be nice for reading

Describe the solution you'd like
add a method that change the color , or the background color

Describe alternatives you've considered
(consoleColor "1") (4Bits console colors)
(consoleBackgroundColor "2") (4 bits console colors)
Additional context
Mmmh, for better interactions beetwen user and program