4548: start of a compiler for a new experimental low-level language

1 files changed, 32 insertions, 0 deletions

diff --git a/transect/compiler5 b/transect/compiler5
new file mode 100644
index 00000000..aeb857f4
--- /dev/null
+++ b/transect/compiler5
@@ -0,0 +1,32 @@
+== Goal
+
+A memory-safe language with a simple translator to x86 that can be feasibly written in x86.
+
+== Definitions of terms
+
+Memory-safe: it should be impossible to:
+  a) create a pointer out of arbitrary data, or
+  b) to access heap memory after it's been freed.
+
+Simple: do all the work in a 2-pass translator:
+  Pass 1: check each instruction's types in isolation.
+  Pass 2: emit code for each instruction in isolation.
+
+== types
+
+int
+char
+(address _ t), t ∋ {stack, heap, global}
+(array _ t), t ∋ {stack, heap, global}
+
+stack addresses can't be copied to heap or global
+heap addresses can't be copied [1]
+global addresses you're free to use anywhere
+
+[1] (address _ heap) can't be copied or stored, can't be part of a type or
+choice. Only thing you can do with it is access it from the register you wrote
+it to. And even that not past a call instruction. Important detail: `free()`
+is a call. So an address to something on the heap can never be invalid if the
+program type-checks.
+
+<reg x> : (address T m) <- advance <reg/mem> : (array T m), <reg offset> : (index T)