Merge pull request #39 from rafalgrodzinski/26-update-readme

Update readme
This commit is contained in:
Rafał
2025-08-14 17:33:29 +09:00
committed by GitHub
5 changed files with 295 additions and 31 deletions

View File

@@ -1,42 +1,64 @@
# Bits Runner Builder
Welcome to Bits Runner Builder! Compiler for the Bits Runner Code (BRC) language 🤘
## Quick links
- [BRC Language Reference](Reference.md)
- [Detailed Syntax](Syntax.md)
## Overview
Bits Runner Builder (brb) is a compiler for Bits Runner Code (brc) language, which has been designed for the Bits Runner Builder operating system. It aims to be a low-level language, which can be a replacement for C while providing a revised syntax and a couple of quality of life improvement. It's a simple system programming language, so no class hierarchies, templates, or other unnecessary fluff.
Bits Runner Builder is a compiler for Bits Runner Code (brc) language, which has been designed for the [Bits Runner](https://github.com/rafalgrodzinski/bits-runner) operating system. It aims to be an opinionated, low-level language, a sort of improved C while providing a revised syntax and a couple of quality of life improvement. It's a simple system programming language, so no class hierarchies, templates, or other unnecessary fluff.
It has been been built on top of LLVM.
It has been been built with LLVM so it should be fairly performant. Keep in mind that it is still work in progress so not everything is finished and there is still probably plenty of bugs and gremlins hiding around 🙈
## Show me the code!
## Main features
BRC allows for low-level system programming, so one of the main features is a seamless support for embeded assembly, pointers mainipulation, and explicit data handling. For this reason data types have explicit byte-sizes, there is no runtime and the memory is manually managed.
### Comments
Like in C, comments can specified using either `\\` which will run until the end of the line or through `/* */` block. However, unlike C, the `/* bla bla /* bla */ */` comments can be also embeded inside each other.
### Literals
**Number literals** can be specified as decimal, hexadecimal, and binary numbers. Digits can be separated by an '_' but it cannot be the first or the last character (otherwise it will get interpreted as and identifier).
The language aims to be simple, easy to reason about, and predictable. Because of this there a class-like features, but no inheritance. Composition is much better anyway and doesn't lead to incomprehensible codebases (did I mention that it's opinionated?).
## Examples
```
// Valid examples:
1024
1_024
1.245
// Basic hello world
//
@extern putchar fun: character u32 -> u32
1_000.
main fun -> u32
text data<u8> <- "Hello, world!\n"
rep i u32 <- 0, text[i] != 0:
putchar(text[i])
i <- i + 1
;
0xffa
0xffaa_42bb
0b1101
0b1010_0101
// Invalid examples:
_100
1000_.100
0x_fa
0b10_
_0b1101
ret 0
;
```
### Control flow
## But why?
The idea was to build the whole computing environment from scratch which can be its own thing. Many project of this kind try to be sort of recoding of C/Unix, but this is not the point in the case. This project doesn't aim at compatibility so it may hapilly break things in order to make things simpler, more modern, or just different.
### Functions
It's mostly a learning opportunity and a bit of fun, but maybe you can find some bits of interesting knowledge for your own project.
## Quick Start
Make sure that you have cmake, llvm, and lld installed on your system.
```
cmake -B build
cmake --build build --config Release
// or
cmake --build build --config Debug
```
You'll then be able to finde the executable under `build/brb`.
There are also "Build (Debug)" and "Clean" tasks specified for VSCode. There is also a launch configuartion, which you can launch by pressing F5 will will then build and start debugging using command `brb -v samples/test.brc`. You'll need to have "LLDB DAP" extension installed in VSCode.
## Samples
Hello World
[samples/hello.brc](samples/hello.brc)
Fibonaci Numbers
[samples/fib.brc]()
#### How to build the samples
```
brb samples/hello.brc
cc -o hello hello.o
```

240
Reference.md Normal file
View File

@@ -0,0 +1,240 @@
# BRC Language Reference
## Overview
Bits Runner Code (BRC) borrows a lot of concepts and syntax from C, but in a slightly modified way. The idea is to use familiar concept in a simplified way, avoiding usage of unnecessary fluff and just to make the code simpler and shorter, while avoiding any unambigouity.
Semicolons are not placed at the end of statements, but instead they delimit blocks, such as body of a function or a loop. There are no curly brackets. They are not necessary if you, for example, declare an external method or have an `if` expression in a single line. Round brackets `()` are also not ncessary in most of the cases, for example when defining a function or evaluating a condition, but are required for function calls. New lines also play important role and may be required or invalid, depending on the context.
Single equal sign `=` denotes comparion and instead left arrow `<-` is used as an assign symbol.
Source code is grouped into named modules, each module can be compromised of number of files. There is no separate header file, instead prefix `@pub` if attached to externally visible symbols.
## Language Elements
- Comments (`//`, `/* */`)
- Literals (`123`, `0xa2`, `0b0101`, `3.14`, `"Hello"`, `'!'`, `true`, `false`)
- Operators (`+`, `-`, `*`, `/`, `%`, `<-`, `<`, `<=`, `>`, `>=`, `=`, `!=`)
- Variables (`u8`, `u32`, `s8`, `s32`, `r32`, `data`, `blob`)
- Functions (`fun`)
- Raw Functions (`raw`)
- Conditional Expressions (`if`, `else`)
- Loops (`rep`)
## Comments
Like in C, comments can specified using either `\\` which will run until the end of the line or through `/* */` block. However, unlike C, the `/* bla bla /* bla */ */` comments can be also embeded inside each other.
```
// this is a main function
main fun -> u32
/*
num1 <- 2 + 5
/* num1 <- 4 * num1 */
// num1 <- 5 * num1
*/
ret 0
;
```
## Literals
**Number literals** can be specified as decimal, hexadecimal, and binary numbers. Digits can be separated by an '_' but it cannot be the first or the last character (otherwise it will get interpreted as an identifier).
```
// Valid examples:
1024
1_024
1.245
1_000.
0xffa
0xffaa_42bb
0b1101
0b1010_0101
// Invalid examples:
_100
1000_.100
0x_fa
0b10_
_0b1101
```
**Text literals** can be specified either as an implicitly zero terminated string, or as a single character. Strings are converted into arrays. Characters can be also backslash '\' escaped, just like in C.
```
// Examples
"Hello world"
"Hello world\0" // in this case, the final zero is not appended
'H'
'!'
'\n'
// Escape sequences
'\b' // backspace
'\n' // new line
'\t' // tab
'\\' // backslash
'\'' // single quotation mark
'\"' // double quotiation mark
'\0' // 0 (as in integer 0)
```
**Boolean literals** can also be specified using `true` or `false` keywords. There is no implicit conversion from integer to boolean and vice-versa.
## Operators
All the standard operators, such as `+`, `-`, `*`, `/`, `%` are available. The biggest difference is that the assignment uses the left arrow `<-` and thus a comparison can be done through a single equal sign `=`.
```
+ // addition
- // subtraction
* // multiplication
/ // division
% // division reminder
= // is equal
!= // is not equal
< // is less than
<= // is less or equal
=> // is greater than
> // is gearter or equal
<- // assignment
( ) // precdence
```
## Variables
Variables are specified by first providing the name and then the type. There is also an optional initializer.
```
bytesInKilobyte u32 <- 1_024
text data<u8> <- "Hello world!"
pi r32 <- 3.14
```
**Simple variables**
There are standard float and integer types available, but unlike in C, you have to be explicit about their size and signiness. You can only perform `=` and `!=` operations on booleans. There is no `void` type or an equivalent.
```
u8 // unsigned integer, 8 bits
u32 // unsigned integer, 32 bits
s8 // signed integer, 8 bits
s32 // signed integer, 32 bits
r32 // floating point (real), 32 bits
bool // true or false
```
**Data variables** or arrays, as known in C. They are a sequence of static length or elements of the same type. Length has to be specified either explicitly or through and initializer.
```
text data<u8> <- "Hello world!"
fibonaciNumbers<u32, 4> <- [1, 1, 2, 3, 5, 8] // Anything past the first 4 numbers will be ignored
```
**Blob variables**, otherwise known as structures. Composite types which we can specify by ourselves. The usage is fairly smillar as in C. Semicolon and new line are required in the definition.
```
user blob
age u32
name data<u8, 10>
isActive bool
;
bob user
bob.age <- 18
bob.name <- "Bob"
bob.isActive <- true
```
## Functions
Functions in BRC work just like in C. You can specify an optional list of arguments and a return type. Calls require usage of round brackets. Colon should be omitted if there are no arguments. Arrow has to be on the same line as the return type.
```
// Valid examples
main fun -> u32
ret 0
;
addNums fun: num1 s32, s32 -> s32
ret num1 + num2
;
addNums fun:
num1 s32,
num2 s32
-> s32
ret num1 + num2
;
addNums(5, 4)
logInUser fun: user User
// do some internet stuff 📡
;
logInUser(bob)
explodeEverything fun
// Do a boom! 💥
;
explodeEverything()
// Invalid examples
addNums num1 s32, num2 s32 -> s32
[..]
addNums: num1 s32
,num2 s32 -> s32
[..]
addNums: num1 s32, num2 s32 ->
s32
[..]
```
## Raw Functions
A unique feature of BRC is a seamless use of inline assembly. Raw functions can be used just like normal functions, altoght there is a couple of limitations and they require so called constraints to be specified. It's the same as in gcc or clang, but they are specified as a single string instead of splitting them into input, output, and clobbers. Some more information can be found here [A Practical Guide to GCC Inline Assembly](https://blog.alex.balgavy.eu/a-practical-guide-to-gcc-inline-assembly/). Intel syntax is used for the assembly.
```
rawAdd raw<"=r,r,r">: num1 u32, num2 u32 -> u32
add $1, $2
mov $0, $1
;
// later on
result u32 <- rawAdd(5, 4)
```
## Conditional Expressions
If-Else statements can be written on a single or multiple lines and are an expression, which allows them to return values.
```
isValid bool <- if count = 0: doForEmpty() else doForCount(count)
if numer > 0:
doStuff(number)
else
fatalError()
;
if featureFlag:
// Special case ⚰️
;
if hasCompleted: exit()
if processedElementsCount < 10: print("Success) else
print("Failure")
processFailure(processedElementsCount)
;
```
## Loops
C-style for, while, and do-while are all combined into a single `rep` loop. The format is `rep init, pre-condition, post-condition`. `init` allows to setup a counter, pre-condition is evaluated before and post after each loop. Each part is optional, but if you include post-condition, pre-condition must also be include. Body can be specified on the same line as the loop, in which case the final semicolon should not be included.
```
// infinite loop
rep: doStuff()
// do things ten times
rep i u32 <- 0, i < 10:
doStuff(i)
i <- i + 1
;
// do things at least once
rep i u32 <- 0, true, i < someValue:
doStuff(i)
;
```

View File

@@ -1,6 +1,6 @@
# Detailed Syntax
This documents specifies what is the allowed syntax for statements and expressions!
This documents specifies what is the allowed syntax for statements and expressions 🤓
### Symbols used
`?` 0 or 1 instances

2
lines_count.sh Executable file
View File

@@ -0,0 +1,2 @@
#!/bin/bash
find . \( -name "*.h" -o -name "*.cpp" \) -print0 | xargs -0 wc -l

View File

@@ -253,8 +253,8 @@ shared_ptr<Statement> Parser::matchStatementFunction() {
Parsee::groupParsee(
ParseeGroup(
{
Parsee::tokenParsee(TokenKind::RIGHT_ARROW, true, false, false),
Parsee::tokenParsee(TokenKind::NEW_LINE, false, false, false),
Parsee::tokenParsee(TokenKind::RIGHT_ARROW, true, false, false),
Parsee::valueTypeParsee(true, true, true)
}
), false, true, false