Merge pull request #246 from sysprog21/improve-perf

Minor performance improvements

Author: jserv

src/defs.h

@@ -79,6 +79,7 @@ typedef struct arena_block {
 typedef struct {
     arena_block_t *head;
     int total_bytes; /* Track total allocation for profiling */
+    int block_size;  /* Default block size for new blocks */
 } arena_t;
 
 /* string-based hash map definitions */

src/globals.c

@@ -144,6 +144,8 @@ arena_t *arena_init(int initial_capacity)
     }
     arena->head = arena_block_create(initial_capacity);
     arena->total_bytes = initial_capacity;
+    /* Use the initial capacity as the default block size for future growth. */
+    arena->block_size = initial_capacity;
     return arena;
 }
 
@@ -166,9 +168,12 @@ void *arena_alloc(arena_t *arena, int size)
     size = (size + PTR_SIZE - 1) & ~(PTR_SIZE - 1);
 
     if (!arena->head || arena->head->offset + size > arena->head->capacity) {
-        /* Need a new block: choose capacity = max(DEFAULT_ARENA_SIZE, size) */
-        int new_capacity =
-            (size > DEFAULT_ARENA_SIZE ? size : DEFAULT_ARENA_SIZE);
+        /* Need a new block: choose capacity = max(DEFAULT_ARENA_SIZE,
+         * arena->block_size, size) */
+        int base =
+            (arena->block_size > DEFAULT_ARENA_SIZE ? arena->block_size
+                                                    : DEFAULT_ARENA_SIZE);
+        int new_capacity = (size > base ? size : base);
         arena_block_t *new_block = arena_block_create(new_capacity);
         new_block->next = arena->head;
         arena->head = new_block;

src/lexer.c

@@ -235,20 +235,23 @@ bool is_numeric(char buffer[])
 
 void skip_whitespace(void)
 {
+    int pos = SOURCE->size;
     while (true) {
-        if (is_linebreak(next_char)) {
-            SOURCE->size += 2;
-            next_char = SOURCE->elements[SOURCE->size];
+        /* Handle backslash-newline (line continuation) using local pos */
+        if (next_char == '\\' && SOURCE->elements[pos + 1] == '\n') {
+            pos += 2;
+            next_char = SOURCE->elements[pos];
             continue;
         }
         if (is_whitespace(next_char) ||
             (skip_newline && is_newline(next_char))) {
-            SOURCE->size++;
-            next_char = SOURCE->elements[SOURCE->size];
+            pos++;
+            next_char = SOURCE->elements[pos];
             continue;
         }
         break;
     }
+    SOURCE->size = pos;
 }
 
 char read_char(bool is_skip_space)
@@ -296,27 +299,42 @@ token_t lex_token_internal(bool aliasing)
         /* C-style comments */
         if (next_char == '*') {
             /* in a comment, skip until end */
+            int pos = SOURCE->size;
             do {
-                read_char(false);
+                /* advance one char */
+                pos++;
+                next_char = SOURCE->elements[pos];
                 if (next_char == '*') {
-                    read_char(false);
+                    /* look ahead */
+                    pos++;
+                    next_char = SOURCE->elements[pos];
                     if (next_char == '/') {
-                        read_char(true);
+                        /* consume closing '/', then commit and skip trailing
+                         * whitespaces
+                         */
+                        pos++;
+                        next_char = SOURCE->elements[pos];
+                        SOURCE->size = pos;
+                        skip_whitespace();
                         return lex_token_internal(aliasing);
                     }
                 }
             } while (next_char);
 
+            SOURCE->size = pos;
             if (!next_char)
                 error("Unenclosed C-style comment");
             return lex_token_internal(aliasing);
         }
 
         /* C++-style comments */
         if (next_char == '/') {
+            int pos = SOURCE->size;
             do {
-                read_char(false);
+                pos++;
+                next_char = SOURCE->elements[pos];
             } while (next_char && !is_newline(next_char));
+            SOURCE->size = pos;
             return lex_token_internal(aliasing);
         }