Parser optimizations.
- parser is now non-recursive (i.e. parse function is not called recursively in any case);
- byte-code is now more compact:
- constants are now not immediately dumped upon occurence, but later - where necessary;
- assignments are combined with unary / binary operations;
- binary operations are encoded more compactly in many cases;
- byte-code arrays are now allocated separately for each scope (so, GC of the scopes now becomes possible);
- byte-code is dumped directly into corresponding byte-code arrays:
- linked lists of op_meta are not now used for main code of a scope.
JerryScript-DCO-1.0-Signed-off-by: Ruben Ayrapetyan r.ayrapetyan@samsung.com
JerryScript-DCO-1.0-Signed-off-by: Andrey Shitov a.shitov@samsung.com
This commit is contained in:
committed by
Ruben Ayrapetyan
parent
b1de93abd6
commit
50d124bfc3
@@ -31,7 +31,7 @@ typedef struct
|
||||
lit_id_hash_table *lit_id_hash_table_init (uint8_t *, size_t, size_t, size_t);
|
||||
size_t lit_id_hash_table_get_size_for_table (size_t, size_t);
|
||||
void lit_id_hash_table_free (lit_id_hash_table *);
|
||||
void lit_id_hash_table_insert (lit_id_hash_table *, vm_idx_t, vm_instr_counter_t, lit_cpointer_t);
|
||||
vm_idx_t lit_id_hash_table_insert (lit_id_hash_table *,vm_instr_counter_t, lit_cpointer_t);
|
||||
lit_cpointer_t lit_id_hash_table_lookup (lit_id_hash_table *, vm_idx_t, vm_instr_counter_t);
|
||||
uint32_t lit_id_hash_table_dump_for_snapshot (uint8_t *, size_t, size_t *, lit_id_hash_table *,
|
||||
const lit_mem_to_snapshot_id_map_entry_t *, uint32_t, vm_instr_counter_t);
|
||||
|
||||
Reference in New Issue
Block a user