Andrew Dunstan <andrew@dunslane.net> writes:
> well, that's what we have a buildfarm for.
It looks like this problem is in setPathObject:
setPathObject(JsonbIterator **it, Datum *path_elems, bool *path_nulls,
int path_len, JsonbParseState **st, int level,
Jsonb *newval, uint32 npairs, bool create)
{
JsonbValue v;
int i;
JsonbValue k;
bool done = false;
if (level >= path_len || path_nulls[level])
done = true;
/* empty object is a special case for create */
if ((npairs == 0) && create && (level == path_len - 1))
{
JsonbValue new = k;
new.val.string.len = VARSIZE_ANY_EXHDR(path_elems[level]);
new.val.string.val = VARDATA_ANY(path_elems[level]);
(void) pushJsonbValue(st, WJB_KEY, &new);
Since "k" is undefined at this point, initializing "new" that way is pure
hogwash. I'm surprised gcc doesn't complain about it.
(BTW, could I lobby to not use "new" as a variable name? lldb gets
confused, probably because "new" is a C++ keyword.)
regards, tom lane