The ASSERT_CMP macros like ASSERT_CMPINT use the arguments that are being compared directly in the error message.

For example, the ASSERT_CMPINT approximately looks like:

#define ASSERT_CMPINT(a, b)                  \
if (a != b) {                                \
   printf("assert failure: %d != %d", a, b); \
   abort();                                  \ 
} 

The caveat is that if a or b are function calls (that aren't idempotent), the error message might be misleading.

Here's an example of one such case:

ASSERT_CMPINT (
    0, ==, bson_json_reader_read (reader, &bson_out, &error));

Since b is substituted for that function call, that function gets called twice, and the error message could produce something confusing like "0 != 0".