43 Commits

Author	SHA1	Message	Date
haxala1r	c9694af826	vm and compiler: removed automatic currying, and several other modifications to the language	2026-05-10 17:23:42 +03:00
haxala1r	947d2274bb	vm: modified StoreLocal and StoreGlobal logic to be more consistent with the rest of the VM, ... and modified the emit module to emit a Pop instruction after every top-level expression. This change was required because the semantics of the language are pretty clear. Every expression evaluates to something - meaning that, in the corresponding bytecode, every expression must have exactly a +1 effect on the data stack. I.e. every expression, when its corresponding bytecode is evaluated, has the effect of pushing something to the stack. For values that are not used by another expression, this value must be immediately popped. Some optimizations could target this area. For example, for top-level expressions, it is obvious to the compiler that their values will not be used - hence the compiler can use optimized versions of some instructions (like StoreLocal and StoreGlobal) to simply never leave the value on the stack, thus saving an extra Pop instruction (good for performance and code size). Same thing applies in function bodies, letrec/let/begin bodies, where expressions whose values are never used may appear. It may also make sense to introduce registers to the VM, for the purposes of parameter passing (such that up to a predetermined number of parameters are progressively passed through registers instead of pushed to the stack). This would pair well with eliminating unnecessary currying in the byte code.	2026-04-26 01:20:05 +03:00
haxala1r	d846046c4a	Big changes: ... Compiler and VM are now working well enough to support larger programs. A test program is included and executed immediately upon running the executable. A more feature complete read-eval-print loop is planned.	2026-04-26 00:55:43 +03:00
haxala1r	0925b44ef7	vm: got the VM to finally actually work	2026-04-25 22:48:06 +03:00
haxala1r	4716c71b15	compiler: modify the Emit module to actually produce a vm_state object directly executable by the VM	2026-04-25 21:04:49 +03:00
haxala1r	2b02740e68	compiler: add the initial draft for compiling into byte code with backpatching	2026-04-25 20:46:28 +03:00
haxala1r	190ec94e14	compiler: make space for compiling into the VM bytecode	2026-04-25 20:44:37 +03:00
haxala1r	757df161d1	scope analysis: created another type to statically eliminate one (im)possible error case	2026-04-05 23:14:29 +03:00
haxala1r	4792a296d3	interpreter: minor fixes, changes to the main binary to actually interpret some source code	2026-03-22 17:23:49 +03:00
haxala1r	cb7e3f8622	scope_analysis: changed convert to return the global symbol table as well as the program	2026-03-01 16:24:45 +03:00
haxala1r	fe26b6c2b3	scope_analysis & core_ast: added functions to convert directly from source for convenience	2026-02-28 18:59:37 +03:00
haxala1r	a95a676cb2	scope_analysis: small logic fix	2026-02-28 18:41:02 +03:00
haxala1r	24db34db62	scope_analysis: added support for deferred computation	2026-02-15 15:09:59 +03:00
haxala1r	8d301a6fc2	scope_analysis: fix the handling of Lambda forms	2026-02-12 18:32:36 +03:00
haxala1r	81dfc07867	compiler: added my first attempt at a scope analysis pass	2026-02-11 23:40:53 +03:00
haxala1r	bd6acf89e0	util: separated the monadic traverse into a utility module	2026-02-11 23:40:17 +03:00
haxala1r	a250d96c63	core_ast: remove letrec comment	2026-02-04 21:52:16 +03:00
haxala1r	3a3bf2c674	core_ast: removed letrec. we now treat it as let + set.	2026-02-02 22:36:55 +03:00
haxala1r	7402a688c7	compiler: remove unused files	2026-02-02 20:43:13 +03:00
haxala1r	ed4f49311c	syntactic_ast: unwrapped the central GADT into several related types. ... core_ast: related to the above, reorganized core_ast to use the new syntactic ast	2026-02-02 20:32:15 +03:00
haxala1r	bd907fe69a	minor changes	2026-02-02 19:53:08 +03:00
haxala1r	d7729571ea	core_ast: modify the core ast to use unary functions	2026-01-16 22:34:35 +03:00
haxala1r	12b347de38	general: removed containers dependency	2026-01-14 21:40:21 +03:00
haxala1r	11625c88c6	core_ast: add of_sexpr	2026-01-14 12:28:22 +03:00
haxala1r	d41d8e5fbe	core_ast: add initial draft for the core ast, and a conversion function from the syntactic ast	2026-01-07 20:02:10 +03:00
haxala1r	ec56d76fb3	syntactic_ast: modify the phantom types to have concrete constructors, as the ocaml compiler cannot prove them distinct otherwise	2026-01-07 20:00:11 +03:00
haxala1r	be81061895	syntactic_ast: fix issue in set! parsing, add set! to the test case.	2026-01-07 18:18:38 +03:00
haxala1r	e25b6b0b10	syntactic_ast: fix issue in cond parsing	2026-01-07 18:17:24 +03:00
haxala1r	0d731f29b3	syntactic_ast: added a node for set! expressions	2026-01-07 18:13:55 +03:00
haxala1r	54c48ddf0e	syntactic_ast: add node for if expressions	2026-01-07 18:03:28 +03:00
haxala1r	edc9d8b9e6	syntactic_ast: added parsing and pretty printing for cond expressions	2026-01-04 21:38:04 +03:00
haxala1r	e95a115acf	syntactic_ast: Added proper handling for Let and Letrec	2026-01-03 21:35:27 +03:00
haxala1r	9fb29afc3e	syntactic_ast: Modified letrec forms to also accept a body	2026-01-03 20:49:17 +03:00
haxala1r	17e533dbb8	Modified the GADT to distinguish bodies	2026-01-03 20:43:11 +03:00
haxala1r	6d95977324	ast: improved the implementation of syntactic_ast ... debug: added debug functions for the syntactic_ast module Modified the compiler executable to test a little bit. todo: Some nodes of the syntactic ast are not yet emitted,	2026-01-03 18:30:47 +03:00
haxala1r	cb94372f29	new ast: added a new syntactic_ast module that provides a more structured AST as a GADT	2026-01-03 16:44:00 +03:00
haxala1r	bc7ca0fa2d	new library: added containers as a dependency.	2026-01-01 13:36:09 +03:00
haxala1r	9981c222eb	Added desugar phase for converting cond into a chain of if's	2025-12-28 21:04:45 +03:00
haxala1r	624d420afe	Separated mutually recursive definition of desugar phases ... They do not need to be mutually recursive at all. At first I thought they would have to be, because I thought it would be best if each phase completely desugared the bodies of the expressions they handle, but now I see that would probably lead to an infinite loop. There is simply no reason to couple them so tightly.	2025-12-28 21:04:45 +03:00
haxala1r	d80caadc85	Added gensym, and phases in the desugar module for logical and & or	2025-12-28 21:04:45 +03:00
haxala1r	be0d479665	Added desugar steps to add (begin) wrappers around long lambda bodies, and rewrite internal (define) forms into letrec forms	2025-12-28 21:04:45 +03:00
haxala1r	4f1ad54596	Started work on the compiler, with a desugar phase	2025-12-28 21:04:45 +03:00
haxala1r	adbf083c3d	Reorganized project	2025-12-28 21:04:45 +03:00