feat/8-ast-datastructures #11

alexander.nutz · 2025-10-20T21:20:36+02:00

alexander.nutz commented

2025-10-20 21:20:36 +02:00

Example usage:

  click_ast_k_unary* x = click_ast_new(header->_obj._arena, click_ast_k_unary);
  click_ast* xx = &x->header;

  click_ast_dup(x);
  click_ast_drop(x);
  click_dup(click_ast, xx);
  click_drop(click_ast, xx);
  click_ast_drop(x);

closes #8

Example usage: ```c click_ast_k_unary* x = click_ast_new(header->_obj._arena, click_ast_k_unary); click_ast* xx = &x->header; click_ast_dup(x); click_ast_drop(x); click_dup(click_ast, xx); click_drop(click_ast, xx); click_ast_drop(x); ``` closes #8

alexander.nutz added 2 commits

2025-10-20 21:20:36 +02:00

feat(libclick): ref-counted object system 646878e0e2

feat(libclick): ast datastructure e70a6b4002

thacuber2a03 self-assigned this

2025-10-20 21:21:39 +02:00

thacuber2a03 reviewed

2025-10-20 21:24:55 +02:00

include/click/click.h

					
				@ -50,2 +69,4 @@

				typedef int click_exitcode;

				#define click_borrow /* borrow ownership for lifetime */

				#define click_move   /* permanently move ownership */

what do you plan on doing here?

just improve code readability; also there are ownership specifiers in c23 I thnk? didn't look at it yet

alexander.nutz marked this conversation as resolved

thacuber2a03 approved these changes

2025-10-20 23:11:22 +02:00

thacuber2a03 left a comment

I'm merging all of this but I'm going to take the time to document it all as really only you can read it

include/click/click.h

					
				@ -199,0 +292,4 @@

				  } while (0)

				#define click_dup_impl(/* borrow */ obj) \

				  click_debug_assert(((click_obj_header*) (obj))->_rc++ != 0);

what does it mean for _rc to be -1

what does it mean for `_rc` to be -1

this just makes sure we aren't overflowing refcount

but actually that should be a pre increment

sob

just patch it into main I guess I can't be bothered

sob just patch it into main I guess I can't be bothered

but merge conflicts with the header PR
I'll open an issue instead and we do it after header PR merged

but merge conflicts with the header PR I'll open an issue instead and we do it after header PR merged

alexander.nutz marked this conversation as resolved

include/click/click.h

					
				@ -199,0 +303,4 @@

				#define click_as_obj(t, ptr)                               \

				  ((click_obj_header*) &(                                  \

				       (t*) click_obj_forward_ptr_and_assert(t, (t*) ptr)) \

				       ->_obj)

_obj is only really explained once you get to click_ast.h, might wanna check that

`_obj` is only really explained once you get to `click_ast.h`, might wanna check that

include/click/click_ast.h

					
				@ -0,0 +33,4 @@

				  // this or any of the children had parser error

				  bool contains_parser_error : 1;

				  // this had parser error itself

				  bool is_parser_error : 1;

"this"

(would replace with "this node", assuming everything here is, in fact, the header for a node. granted, I still have no idea what the point of an obj is)

"this" (would replace with "this node", assuming everything here is, in fact, the header for a node. granted, I still have no idea what the point of an obj is)

alexander.nutz marked this conversation as resolved

include/click/click_ast.h

					
				@ -0,0 +49,4 @@

				#define click_ast_header_children(ast_header)                   \

				  ((click_ast**) (((char*) (void*) (click_ast*) (ast_header)) + \

				                  click_ast_kindinfo[(ast_header)->kind].children_offset))

oh my god

alexander.nutz marked this conversation as resolved

libclick/util.c

					
				@ -235,0 +238,4 @@

				void click_check_impl_fail(void)

				{

				  click_panic("click object generational reference fail -> reference was kept longer than object lifetime");

				}