pankhuri(dot)sai(at)gmail(dot)com writes:
> n executing the following query, it is found that no. of actual rows after
> hash join of part and partsupp (158960) is not equal to no. of rows sorted
> after that(158755).
I don't see any bug there. It looks like the mergejoin stopped a bit
short of reading all the output from the sort node, which is unsurprising
behavior for a mergejoin: it can stop as soon as it's exhausted the output
from either input.
regards, tom lane