napi (c not c++) - am i responsible for garbage-collecting external c-buffer passed to napi_create_external_arraybuffer()?

[kaizhu256]

which of the 2 c-code below are correct? or are they both wrong, and its something else?

1. i'm responsible for garbage-collecting cstring passed to napi_create_external_arraybuffer.

void napiFinalize(
    napi_env env,
    void *finalize_data,
    void *finalize_hint
) {
// this function will free <finalize_data>
    free(finalize_data);
}
...
char *cstring = malloc(256);
napi_value result;
napi_create_external_arraybuffer(
    env,
    (void *) cstring,
    256,
    napiFinalize, // explicitly free <cstring> when object is gc'd
    NULL,
    &result
);
return result;

2. nodejs will automatically free external cstring.

char *cstring = malloc(256);
napi_value result;
napi_create_external_arraybuffer(
    env,
    (void *) cstring,
    256,
    NULL, // assume nodejs will automatically free <cstring>
    NULL,
    &result
);
return result;

i'm troubleshooting an addon in win32 environment, that exits silently after a c-call (no segfault or loud warnings). a definitive answer to above might help diagnosing the bug.

[kaizhu256]

nevermind, turns out option1 is correct (user needs to manually garbage collect with free()).

the silent-exit bug was due to something else -- cstring was realloc'd but forgot to set it to new pointer.