The Dryft REPL provides an interactive environment for writing and testing Dryft code snippets without creating files.
Starting the REPL
Launch the REPL by running the compiler without an input file:
You’ll see the REPL prompt:
REPL with Custom Target
You can start the REPL with a specific compilation target:
The x86 backend is not stable in the REPL environment. If you start the REPL with -t elf, you’ll see a warning message. Use the default GCC target for REPL sessions.
REPL Commands
The REPL supports special commands prefixed with a dot (.):
.help or help
Displays the help screen with all available commands.
Dryft repl> .help
.help/help => display this screen
.exit/.quit => leave the REPL, terminating this process
.view => inspect the contents of generated IR from last execution
.view
Inspects the intermediate representation (IR) generated from the last code execution.
Dryft repl> 1 2 +
Dryft repl> .view
.exit or .quit
Exits the REPL and terminates the process.
How the REPL Works
The REPL provides immediate feedback by compiling and executing your code behind the scenes.
Execution Flow
- Input: You enter a Dryft expression or statement
- Wrapping: Your code is automatically wrapped in a
main action with std/io included:
include std/io act: main { YOUR_CODE } ;
- Compilation: The wrapped code is compiled using the selected backend
- Assembly: The IR is assembled into an object file
- Linking: Objects are linked into an executable
- Execution: The executable runs and displays output
Example Session
Dryft repl> 5 3 +
8
Dryft repl> "Hello, Dryft!" print
Hello, Dryft!
Dryft repl> .view
// ... displays generated C code ...
Dryft repl> .exit
REPL Implementation Details
The REPL is implemented in src/main.rs:34-83. Here’s how it works:
Code Wrapping
Every input is wrapped in a standard template (line 63):
let src = format!("include std/io act: main {} ;", input);
Build Steps
The REPL performs the full compilation pipeline for each input:
- Compile: Generates IR using the selected backend
- Stdlib: Compiles standard library (if needed)
- Assemble: Converts IR to object code
- Link: Creates executable
- Interpret: Runs the executable and displays output
Target Specification
The REPL uses the same target system as file compilation. The target determines:
- Which backend generates the IR
- Where intermediate files are stored
- How the code is assembled and linked
- How the final executable is run
Limitations
x86 Backend Instability
The x86 backend has known issues in REPL mode due to how code is injected directly into main. A warning is displayed if you use this backend:
Warning: the x86 backend is not stable in the REPL environment
State Persistence
Each REPL input is compiled and executed independently. Variables and functions defined in one input are not available in subsequent inputs.
Dryft repl> var: x 10 >>x
Dryft repl> x // Error: x is not defined
Output Buffering
Currently, the REPL waits for the executed command to finish before displaying output. This means interactive programs may not work as expected.
Tips for Using the REPL
Use .view frequently to understand how Dryft compiles your code. It’s an excellent learning tool.
Start with simple expressions to get familiar with Dryft’s stack-based syntax before moving to complex programs.
If you encounter errors, check that your target dependencies (like gcc or nasm) are installed and accessible.
Use Cases
- Learning: Experiment with Dryft syntax interactively
- Testing: Quickly test expressions and built-in functions
- Debugging: Use
.view to inspect generated code
- Prototyping: Try out ideas before committing to a file