This is a readable Assembly dialect made of prolog-like pattern matching macros for CLR ILAsm.
Write this:
((write: "Hello, world!"))
Instead of that:
ldstr "Hello, world!"
call void [mscorlib]System.Console::Write (string)
The macrosystem is just a proof of concept, but it works: you can build the code, write your own macros, write your own CLR applications.
<<console_input_output.min>>
<<branching.min>>
<<instructions.min>>
.assembly extern mscorlib {}
.assembly hello
{
.ver 1:0:1:0
}
.module hello.exe
.method static void main() cil managed
{
.maxstack 2
.entrypoint
.locals init (int32 a, int32 b)
((write: "Input positive x: "))
((read line))
((parse it as int32))
((store it as local a))
((check if ((local a)) > ((int32 0)) ))
((if so {{
((write: "It is positive!\n"))
((load int32 2))
((store it as local b))
((write: "x * 2 = "))
((load local a))
((load local b))
((multiply them))
((write int32))
((write line: ""))
}}else{{
((write: "It isn't positive.\n"))
}}
))
((read line))
((return))
}
This piece of code is a valid ILAsm code with MILAsm macros. Not a mock-up.
It reads a number from the console, checks if it's positive, and if it is, returns the double of this number. What's important, you can read the intention from the code yourself even if you don't know any of CLR assembly. The langauge is meant to be both low-level and readable.
Macros are defined in free-form syntax using double square brackets, like this:
[[write $type -> call void [mscorlib]System.Console::Write ($type) ]]
The macro expansion works like this: the part before the -> is a pattern, and the part after the arrow is the substitution. $type is a macro argument. All the words starting from $ are macro arguments. When a pattern occurs in the code, it's being replaced recursively until no recursive patterns are left.
ldstr "Hello!"
((write string))
Becomes:
ldstr "Hello!"
call void [mscorlib]System.Console::Write (string)
Of course, you can multiline the body of a macro like this:
[[write: $text ->
ldstr $text
call void [mscorlib]System.Console::Write (string)
]]
And you can make recursive macros:
[[write: $text ->
ldstr $text
((write string))
]]
And with this new macro, you can turn this:
ldstr "Hello!"
call void [mscorlib]System.Console::Write (string)
To that:
((write: "Hello!"))
When you want to write a macro with the block of other macros or some ILAsm code as a macros variable, use double curly brackets: {{ }}. You can make macros for if or for with that.
Note that apart from the parenthesis, the syntax of all the macros is non-restricted. Macroexpansion uses primitive Prolog-style pattern-matching with atoms being always separated by space, so you can use almost any symbols in your macro definition. Just remember that all identifiers starting with $ are supposed to be macro variables.
And there is one special macro variable: $#. It is a macro expansion for the instance ID. it is intended for branching, so you can make labels like HERE_$#: automatically expanding like HERE_$5: or HERE_$16: or HERE_$192: anew every time the macro expansion is happening.
Another feature is file inclusion dedicated to macros only. You can write your macros definitions in the same file your code is, but you can store them elsewhere and then include them with double corner braces like this:
<<console_input_output.min>>
I didn't want to mess with all CLR infrastructure, so this is not the usual C-style inline. it deliberately works with macros definitions only ignoring all the other text: test code or comments. I'd recommend keeping an extension for these files .min (for "macros include") to avoid mixing them with other code files but this is not an inherent restriction.