…sing in a char/int to mutate

…hp_session functions also expect smart_string instead of char* + int

… zend_string instead of char* + len. IS_BOOL is not a distinct type anymore IS_TRUE/IS_FALSE instead

…ss of zval indirection

…e, and zend_string instead of char+int for zend_lookup_class

…nd msgpack_unserialize_raw to have one level less of dereference

…macro removal PHP 5 -> 7

…sgpack_unpacker_fetch_object and msgpack_base_fetch_object

…ustom object API

… API

…nd_parse_parameters, which returns a zend_string instead of char+int

…ack_serialize

…was being fired when zend_symtable_update succeeded instead of failed

…rocessing refs

…equals IS_INDIRECT

…il we I do a proper refactor of zval allocation for the MSGPACK_UNSERIALIZE_ALLOC_STACK and MSGPACK_UNSERIALIZE_ALLOC_VALUE macros

…pack_serialize_object so it passes an initalized zval to call_user_function_ex

…] instead of *zval[], following the current pattern in ext/standard/var_unserializer.re

…codes, still need to migrate

…end_parse_parameters instead of val+len

…e_property to update a class instead of zend_symtable_update, which would cause undefined behavior

…s to a map that is a reference

…d object as a zval instead of a ulong

…by zend_mangle_propery_name, since we now use zend_update_property instead of zend_symtable_update

…etching a value, fixing the caller

…ize_map_item

…d return false, fixes test 018

…ng a map with a string key

…dling and freeing cleaner

…o msgpack_unserialize_var_destroy causing a segfault

…would object_init_ex and squash the zval ref

…y is 0

…all a method on an object, use call_user_function_ex instead

…tly set php_only using Z_LVAL_P causing it to always be true

…te properties to proper zend_object handlers

…was unused

…val's are still used by non_dtor members

…invalid objects (non-string and 0 length keys)