The document said:
The bson_reinit() function shall be equivalent to calling bson_destroy() and bson_init().
But this demo shows their difference:
test.cc
#include <cassert> #include <cstring> #include <mongoc.h> void log_bson(const bson_t *bson) { char *json = bson_as_json(bson, NULL); puts(json ? json : "[bson_as_json_error]"); if (json) bson_free(json); } void fill(bson_t *p1, bson_t *p2) { bool status; status = bson_append_document_begin(p1, "$set", 4, p2); assert(status); status = bson_append_bool(p2, "foo", 3, true); assert(status); status = bson_append_document_end(p1, p2); assert(status); } #define USE_BSON_REINIT 1 int main() { bson_t bson = BSON_INITIALIZER; bson_t child = BSON_INITIALIZER; fill(&bson, &child); log_bson(&bson); #if USE_BSON_REINIT bson_reinit(&bson); #else bson_destroy(&bson); bson_init(&bson); #endif bson_destroy(&child); bson_init(&child); fill(&child, &bson); log_bson(&child); bson_destroy(&bson); bson_destroy(&child); }
Although outputs are the same when enable and disable USE_BSON_REINIT, valgrind detects a memory leak when USE_BSON_REINIT is true. Something like:
==7558== 128 bytes in 1 blocks are definitely lost in loss record 607 of 663 ==7558== at 0x4C2B0AF: malloc (vg_replace_malloc.c:296) ==7558== by 0x598260B: bson_malloc (in /usr/lib64/libbson-1.0.so.0.0.0) ==7558== by 0x5976EEC: _bson_grow (in /usr/lib64/libbson-1.0.so.0.0.0) ==7558== by 0x5978307: _bson_append_bson_begin (in /usr/lib64/libbson-1.0.so.0.0.0) ==7558== by 0x108C69: fill(_bson_t*, _bson_t*) (test.cc:14) ==7558== by 0x108DD1: main (test.cc:27)
- is related to
-
CDRIVER-811 Clearer Design & document on Invalid bson_t Object
-
- Closed
-