Re-implemented the 'other' literal types: (#91)

Datetime
Time
Date
Size
Duration

URLs are done not as a separate type, but as a literal modifier that
calls CURL's URL validation (loose as it is).

Added an option to turn on / off static binding of calls exposed via
the FFI. If not set, it's on if glibc is off on linux, and off by
default on mac.

Also, for simple assignment ONLY, if the assignment is to a variable
with an explicitly declared type (and that type is a simple literal),
then we will infer the litmod type, picking the first one added
associated with the type.w

Some issue with the dynamic binding test in meson; just always leave static on for now.

doc/reference.md

@@ -533,74 +533,31 @@ BITYPE ::= "void" | "bool" | "i8" | "byte" | "i32" | "char" | "u32" | "int" |
 EOS ::= '\n' | ';' <<or, if followed by a '}' or line comment, then ''>>
 ```
 
-# Builtin methods used by the language
-- $repr
-
 # Features to re-add from old con4m
 - VM save restore
 - Arg parsing
-- Libraries
-- Doc API.
-- Checkpointing
-- Hot loading
-- Finish data types (date, ip, and extra hatrack stuff)
+- Change $result back to result (or alias?)
+- Libraries (but only the ones Chalk needs)
 - Final mile: specs
 - Final mile: params
-- Auto-import standard library.
+- Doc API.
 - Callbacks
-- Len, etc.
+- Make litmods optional when types are declared. 
 
 # Items for afterward
 - Objects
 - Folding
 - Casting
+- Mixed
 - Varargs functions
-- Clean up unused instructions in VM
-- Remove the two-words-per-stack-slot thing; it's not needed anymore.
 - automatic logfd + optional server for log messages
 - REPL
 - Keyword arguments
 - 'maybe' types / nil
 - Aspects (before / after / around pattern(sig) when x) 
-- Casting
-- (Possibly) re-add := literals
 - Threading
-- Mixed
 - Pretty printing w/ type annotations
 - Language server
 - Checks based on PDG
 - Full-program info on unused fields & fns.
 
-# Features removed (considered for adding back in)
-Also, the language accepts ":=", which has the special syntax of
-generating a "=" operator, then tries to automatically determine the
-(non-numeric) literal type automatically by running through a
-prioritized list of literal parsers (currently intended for the
-config-focused literals).
-
-For instance:
-
-```
-now := 11 Jan 2024
-```
-
-Is equal to:
-
-```
-now = "11 Jan 2024"'date
-```
-
-Note that currently, without a literal modifier, "" always assume
-'string', so the following does NOT work:
-
-```
-var now: date
-now = "11 Jan 2024"
-```
-
-This gives a string vs. date type error. `:=` is the thing that says,
-"try to be smart!"
-
-The lexer actually does the conversion, adding an assignment token and the
-`OTHERLIT` token (which maybe we should change to be more descriptive?)
-

include/adts/datetime.h

@@ -0,0 +1,14 @@
+#pragma once
+#include "con4m.h"
+
+typedef struct c4m_date_time_t {
+    struct tm    dt;
+    int64_t     fracsec;
+    unsigned int have_time     : 1;
+    unsigned int have_sec      : 1;
+    unsigned int have_frac_sec : 1;
+    unsigned int have_month    : 1;
+    unsigned int have_year     : 1;
+    unsigned int have_day      : 1;
+    unsigned int have_offset   : 1;
+} c4m_date_time_t;

include/adts/duration.h

@@ -0,0 +1,12 @@
+#pragma once
+#include "con4m.h"
+
+extern c4m_duration_t *c4m_now(void);
+extern c4m_duration_t *c4m_timestamp(void);
+extern c4m_duration_t *c4m_process_cpu(void);
+extern c4m_duration_t *c4m_thread_cpu(void);
+extern c4m_duration_t *c4m_uptime(void);
+extern c4m_duration_t *c4m_program_clock(void);
+extern void            c4m_init_program_timestamp(void);
+extern c4m_duration_t *c4m_duration_diff(c4m_duration_t *,
+                                         c4m_duration_t *);

include/adts/ipaddr.h

@@ -0,0 +1,12 @@
+#pragma once
+#include "con4m.h"
+
+// I realize some of this is redundant, but it's just easier.
+typedef struct {
+    char     addr[sizeof(struct sockaddr_in6)];
+    uint16_t port;
+    int32_t  af;
+} c4m_ipaddr_t;
+
+extern void
+c4m_ipaddr_set_address(c4m_ipaddr_t *obj, c4m_str_t *s, uint16_t port);

include/adts/list.h

@@ -24,3 +24,4 @@ extern bool        c4m_list_contains(c4m_list_t *, c4m_obj_t);
 extern c4m_list_t *c4m_list_copy(c4m_list_t *);
 extern c4m_list_t *c4m_list_shallow_copy(c4m_list_t *);
 extern void        c4m_list_sort(c4m_list_t *, c4m_sort_fn);
+extern void        c4m_list_resize(c4m_list_t *, size_t);

include/adts/string.h

@@ -24,7 +24,7 @@ extern c4m_utf8_t         *c4m_rich(c4m_utf8_t *, c4m_utf8_t *style);
 extern c4m_codepoint_t     c4m_index(const c4m_str_t *, int64_t);
 extern bool                c4m_str_can_coerce_to(c4m_type_t *, c4m_type_t *);
 extern c4m_obj_t           c4m_str_coerce_to(const c4m_str_t *, c4m_type_t *);
-extern c4m_list_t         *c4m_str_xsplit(c4m_str_t *, c4m_str_t *);
+extern c4m_list_t         *c4m_str_split(c4m_str_t *, c4m_str_t *);
 extern struct flexarray_t *c4m_str_fsplit(c4m_str_t *, c4m_str_t *);
 extern bool                c4m_str_starts_with(const c4m_str_t *,
                                                const c4m_str_t *);
@@ -37,7 +37,6 @@ extern c4m_utf32_t        *c4m_title_case(c4m_str_t *);
 extern c4m_str_t          *c4m_str_pad(c4m_str_t *, int64_t);
 extern c4m_utf8_t         *c4m_str_to_hex(c4m_str_t *, bool);
 
-#define c4m_str_split(x, y) c4m_str_xsplit(x, y)
 // This is in richlit.c
 extern c4m_utf8_t *c4m_rich_lit(char *);
 

include/compiler/dt_errors.h

@@ -185,6 +185,12 @@ typedef enum {
     c4m_internal_type_error,
     c4m_err_concrete_index,
     c4m_err_non_dict_index_type,
+    c4m_err_invalid_ip,
+    c4m_err_invalid_dt_spec,
+    c4m_err_invalid_date_spec,
+    c4m_err_invalid_time_spec,
+    c4m_err_invalid_size_lit,
+    c4m_err_invalid_duration_lit,
 #ifdef C4M_DEV
     c4m_err_void_print,
 #endif

include/con4m.h

@@ -1,5 +1,4 @@
 #pragma once
-
 #include "con4m/config.h"
 // Useful options (mainly for dev) are commented out here.
 // The logic below (and into the relevent header files) sets up defaults.
@@ -115,6 +114,9 @@
 
 #include "adts/dict.h"
 #include "adts/set.h"
+#include "adts/ipaddr.h"
+#include "adts/datetime.h"
+#include "adts/duration.h"
 
 #include "core/ffi.h"
 #include "util/watch.h"

include/con4m/base.h

@@ -24,6 +24,7 @@
 #include <dlfcn.h>
 #include <pwd.h>
 #include <dirent.h>
+#include <ctype.h>
 
 #include <sys/select.h>
 #include <sys/types.h>

include/con4m/datatypes.h

@@ -39,6 +39,9 @@ typedef struct hatrack_set_st c4m_set_t;
 #include "compiler/dt_module.h"
 #include "compiler/dt_compile.h"
 
+typedef uint64_t        c4m_size_t;
+typedef struct timespec c4m_duration_t;
+
 typedef c4m_str_t *(*c4m_repr_fn)(c4m_obj_t);
 typedef void (*c4m_marshal_fn)(c4m_obj_t,
                                c4m_stream_t *,

include/core/gc.h

@@ -244,6 +244,7 @@ extern void           c4m_get_heap_bounds(uint64_t *, uint64_t *, uint64_t *);
 extern void           c4m_gc_register_collect_fns(c4m_gc_hook, c4m_gc_hook);
 extern c4m_alloc_hdr *c4m_find_alloc(void *);
 extern bool           c4m_in_heap(void *);
+extern void           c4m_header_gc_bits(uint64_t *, c4m_base_obj_t *);
 
 #ifdef C4M_GC_STATS
 uint64_t c4m_get_alloc_counter();

include/core/init.h

@@ -7,36 +7,36 @@ extern char **c4m_stashed_argv;
 extern char **c4m_stashed_envp;
 
 static inline char **
-c4m_raw_argv()
+c4m_raw_argv(void)
 {
     return c4m_stashed_argv;
 }
 
 static inline char **
-c4m_raw_envp()
+c4m_raw_envp(void)
 {
     return c4m_stashed_envp;
 }
 
-extern c4m_list_t *c4m_get_program_arguments();
-extern c4m_utf8_t *c4m_get_argv0();
+extern c4m_list_t *c4m_get_program_arguments(void);
+extern c4m_utf8_t *c4m_get_argv0(void);
 extern c4m_utf8_t *c4m_get_env(c4m_utf8_t *);
-extern c4m_dict_t *c4m_environment();
+extern c4m_dict_t *c4m_environment(void);
 extern c4m_utf8_t *c4m_path_search(c4m_utf8_t *, c4m_utf8_t *);
-extern c4m_utf8_t *c4m_con4m_root();
-c4m_utf8_t        *c4m_system_module_path();
-
+extern c4m_utf8_t *c4m_con4m_root(void);
+extern c4m_utf8_t *c4m_system_module_path(void);
+extern void        c4m_add_static_symbols(void);
 extern c4m_list_t *con4m_path;
 extern c4m_set_t  *con4m_extensions;
 
 static inline c4m_list_t *
-c4m_get_module_search_path()
+c4m_get_module_search_path(void)
 {
     return con4m_path;
 }
 
 static inline c4m_set_t *
-c4m_get_allowed_file_extensions()
+c4m_get_allowed_file_extensions(void)
 {
     return con4m_extensions;
 }

include/core/literal.h

@@ -1,7 +1,7 @@
 #pragma once
 #include "con4m.h"
 
-extern __uint128_t         c4m_raw_int_parse(char *,
+extern __uint128_t         c4m_raw_int_parse(c4m_utf8_t *,
                                              c4m_compile_error_t *,
                                              bool *);
 extern __uint128_t         c4m_raw_hex_parse(char *,
@@ -15,3 +15,4 @@ extern c4m_compile_error_t c4m_parse_simple_lit(c4m_token_t *,
                                                 c4m_utf8_t **);
 extern c4m_builtin_t       c4m_base_type_from_litmod(c4m_lit_syntax_t,
                                                      c4m_utf8_t *);
+extern bool                c4m_fix_litmod(c4m_token_t *, c4m_pnode_t *);

include/core/object.h

@@ -121,3 +121,8 @@ extern const c4m_vtable_t c4m_callback_vtable;
 extern const c4m_vtable_t c4m_flags_vtable;
 extern const c4m_vtable_t c4m_box_vtable;
 extern const c4m_vtable_t c4m_basic_http_vtable;
+extern const c4m_vtable_t c4m_datetime_vtable;
+extern const c4m_vtable_t c4m_date_vtable;
+extern const c4m_vtable_t c4m_time_vtable;
+extern const c4m_vtable_t c4m_size_vtable;
+extern const c4m_vtable_t c4m_duration_vtable;

include/core/type.h

@@ -315,15 +315,15 @@ c4m_type_typespec()
 }
 
 static inline c4m_type_t *
-c4m_type_ipv4()
+c4m_type_ip()
 {
     return c4m_bi_types[C4M_T_IPV4];
 }
 
 static inline c4m_type_t *
-c4m_type_ipv6()
+c4m_type_ipv4()
 {
-    return c4m_bi_types[C4M_T_IPV6];
+    return c4m_bi_types[C4M_T_IPV4];
 }
 
 static inline c4m_type_t *

include/io/http.h

@@ -60,6 +60,14 @@ extern c4m_basic_http_response_t *_c4m_http_upload(c4m_str_t *,
                                                    c4m_buf_t *,
                                                    ...);
 
+static inline bool
+c4m_validate_url(c4m_utf8_t *candidate)
+{
+    CURLU *handle = curl_url();
+
+    return curl_url_set(handle, CURLUPART_URL, candidate->data, 0) == CURLUE_OK;
+}
+
 static inline bool
 c4m_http_op_succeded(c4m_basic_http_response_t *op)
 {

include/util/wrappers.h

@@ -3,9 +3,9 @@
 #include "con4m.h"
 
 extern c4m_utf32_t *c4m_wrapper_join(c4m_list_t *, const c4m_str_t *);
-extern c4m_str_t   *c4m_wrapper_hostname();
-extern c4m_str_t   *c4m_wrapper_os();
-extern c4m_str_t   *c4m_wrapper_arch();
+extern c4m_str_t   *c4m_wrapper_hostname(void);
+extern c4m_str_t   *c4m_wrapper_os(void);
+extern c4m_str_t   *c4m_wrapper_arch(void);
 extern c4m_str_t   *c4m_wrapper_repr(c4m_obj_t);
 extern c4m_str_t   *c4m_wrapper_to_str(c4m_obj_t);
 extern void         c4m_snap_column(c4m_grid_t *, int64_t);

meson.build

@@ -6,7 +6,6 @@ project(
     license: 'MIT',
 )
 
-
 cc = meson.get_compiler('c')
 incdir = include_directories('include')
 using_osx = false
@@ -65,6 +64,12 @@ if get_option('forkless_tests').enabled()
     c_args = c_args + ['-DC4M_TEST_WITHOUT_FORK']
 endif
 
+
+
+if not get_option('static_ffi_binding').disabled()
+  c_args = c_args + ['-DC4M_STATIC_FFI_BINDING']
+endif  
+
 backtrace = cc.find_library('backtrace', required: false)
 
 if backtrace.found()
@@ -208,6 +213,9 @@ c4m_adts = [
     'src/adts/mixed.c',
     'src/adts/tuple.c',
     'src/adts/ipaddr.c',
+    'src/adts/datetime.c',
+    'src/adts/size.c',
+    'src/adts/duration.c',
     'src/adts/callback.c',
     'src/adts/streams.c',
     'src/adts/flags.c',
@@ -328,6 +336,7 @@ test_src = [
 
 threads = dependency('threads')
 math = cc.find_library('m', required: false)
+
 ffi = cc.find_library(
     'ffi',
     required: true

meson.options

@@ -127,4 +127,11 @@ option(
     type: 'feature',
     value: 'auto',
     description: 'Run test suite without forking',
+)
+
+option(
+    'static_ffi_binding',
+    type: 'feature',
+    value: 'auto',
+    description: 'Whether to statically bind native builtins in std library',
 )