Using Lobster from the command line

Basic usage

If you have a file helloworld.lobster that contains

print "Hello, World!"

then running it like so will compile and run it:

bin/lobster helloworld.lobster

Command line options

Format: lobster [ OPTIONS ] [ FILE ] [ -- ARGS ]

FILE : main Lobster file to compile & run.
-- ARGS pass args to the running Lobster program, available from command_line_arguments(). Must be last on the command-line.
--pak : generates a pakfile (currently always called "default.lpak") in the same folder as the .lobster file it reads, and doesn't run the program afterwards. If you run lobster with no arguments at all, it will try to load "default.lpak" from the same folder it resides in. Thus distributing programs created in lobster is as simple as packaging up the lobster executable with a pakfile. The pakfile contains the bytecode, and any data files you have specified with the pakfile keyword, see “Distributing Lobster programs” in the implementation documentation.
--cpp : compiles to a .cpp file. Please make sure to follow implementation on how to use this feature. Useful if you’ve created something in Lobster that could use a bit more speed, for a shipping build. Not recommend to be used during development.
--import RELDIR specifies a dir relative to FILE from which import (and pakfile) statements can be resolved, or any loading the running program does. Does the same as adding import from "RELDIR" to the top of your program.
--runtime-no-asserts : Compile with asserts off, don't use unless you know what you're doing.
--runtime-asserts : Compile with asserts on (default).
--runtime-stack-traces : Compile with asserts on + line info + function tracking for better runtime errors (stack traces) and profiling. Small effect on speed.
--runtime-debug : Same as verbose, but now tones inlining way down to get better stack traces. Larger effect on speed.
--runtime-debug-dump: In addition, creates a memory dump file on error.
--runtime-debugger: Automatically breaks into the graphical debugger on errors.
--main MAIN : after compiling FILE, if present, compile this file and run it. This is useful when running Lobster from an editor, where you may be editing a module that is not the main file, would like to see that it compiles by itself, but to run it need to be launching the main Lobster program it belongs to (e.g. main.lobster in the same folder).
--wait : makes the compiler wait for commandline input before it exits. Useful in Windows batch files.
--noconsole : Close console window (Windows)
--verbose : verbose mode, outputs additional stats about the program being compiled/run. --debug outputs even more, only useful for working on the compiler.
--silent : Only output errors.
--full-error : Lobster is able to generate compile time stack traces when the typechecker fails, but those can get long, so by default they get truncated. Use this option to not truncate them, and also dump additional information on free/local variables in the function context.
--gen-builtins-html : dumps a help file of all builtin functions the compiler knows about to builtin_functions_reference.html. --gen-builtins-names dumps a plain text list of functions, useful for adding to syntax highlighting files etc.
--parsedump : dumps internal representations of the program as AST, and --disasm for a readable bytecode dump. Only useful for compiler development or if you are really curious.
--non-interactive-test : Quit after running 1 frame. Useful for running graphical programs as part of a test suite.
--query: Ask the compiler to answer a query about definitions in the program being compiled. When is this mode, the compiler does not try to do a full compilation, but simply tries to answer the query, including ignoring errors or aborting half-way if it has to. This option is for use by an IDE or LSP server implementation. All further args are passed to the query. The first arg is the kind of query (e.g. definition or complete). The second arg is the file this query should take place in (as a relative path, i.e foo/bar.lobster). The 3rd arg is the line where the query should happen. The 4th arg is the identifier the query is about. The rest of the args depend on the query type, if any. For definition a response will be the location of the definition. For complete, it will return a list of possible fields/methods.

Lobster source code may start with a shebang #! so you can embed the command-line in a Lobster script.

Default directories

It's useful to understand the directories lobster uses, both for reading source code files and any data files the program may use:

the root repo directory: This is the main folder that has the default bin modules data docs samples tests folders inside of it.
the auxiliary directory: this is where the main .lobster file being compiled resides.
the directory for writing files: the same as auxiliary on desktop platforms, but often a special directory on mobile platforms.
On Linux additionally it can load files from /usr/share/lobster/ (or whatever path was configured for install by CMake, see DATADIR) if the above paths don't work. This is to allow package managers to install Lobster in the system directories.

Additionally, if any of these folders contains a modules directory, it will load source code from there as well.

Any of the Lobster builtin commands that load data files specify paths relative to either the main or auxiliary directories (and either / or \ may be used as path separators). If you package up a Lobster program for distribution, all these files can be packed into a pakfile, see --pak.

Output

Running lobster may result in a compiler error, that generally look something like this:

mygame.lobster(960): error: unknown identifier: highscor

If compiled correctly, running will give you output from your own print statements, and additionally at some point may cause a runtime error, which can look something like this:

pythtree.lobster(15): VM error: division by zero
in block -> pythtree.lobster(16)
   i = 0
in block -> pythtree.lobster(16)
in function: branch -> pythtree.lobster(29)
   poly = [[1.000000, 0.000000, 0.000000, 0.000000]:xyzw, [-1.000000, ...]:xyzw, ....]
   len = 4
   scale = 0.700000
   max = 11.000000
   n = 0
in block -> pythtree.lobster(29)

This is called a stack trace. Besides the error and the line it happened on, it will show all functions and blocks that called that code, in reverse order, with any local variables and their values. This helps you get an idea where the problem came from and helps in debugging.