Tsearch2 ranking

2 rows of tsvector:

'bar':2 'baz':3 'foo':1
'bar':2 'baz':1 'foo':3

so source text was:

foo bar baz
baz bar foo

ts_query now is 'foo&baz&baz', so both matched.

How can I honor the correct order of the first row and rank it higher?
The position information is there, why doesn't rank() / rank_cd() not
use it?

Is there any way to make a difference between those to rows? Any? Even
if it's a bad hack? I really need it :/

Thanks in advance


--
Regards,
Hannes Dorbath