> Hi. > > Coverity complained about possible dereference null pointer > in *reindex_one_database* function. > That's not really true. > But the logic is unnecessarily complicated.
Hmm, this code looks quite suspect, but I wonder if instead of (what looks more or less like) a straight revert of cc0e7ebd304a as you propose, a better fix wouldn't be to split get_parallel_object_list in two: get_parallel_table_list for the DATABASE and SCHEMA cases, and get_parallel_tabidx_list (or whatever) for the INDEX case. In the first case we just return a list of values, but in the latter case we also meddle with the input list which becomes an output list ...
Sure, I'll try to do it.
Attached is the prototype version v1.
Any chance to push this forward? Is it worth creating a committfest entry?
