On Sun, 28 Aug 2005, Tang Tim Hei wrote: > > > > > ?H????: Stephan Szabo <[EMAIL PROTECTED]> > > ????: 2005/08/27 ?P???? ?U?? 11:25:49 HKT > > ??????: Tang Tim Hei <[EMAIL PROTECTED]> > > ????: pgsql-general@postgresql.org > > ?D??: Re: [GENERAL] A strange problem > > > > On Sat, 27 Aug 2005, Tang Tim Hei wrote: > > > > > Hi, > > > > > I'm new to postgresql. Anytime I type the following command to the > > > database to run, it give me no result record if table 'country' is > > > empty but can get result if 'country' is not empty. Is this so > > > strange? > > > > Not really. You're doing a cartesian join between test.currency and > > test.country. If there are no rows in test.country, there are no rows in > > the output of the from clause. > > > > > select A.* from test.currency A, test.country B where A.curr_cd='USD' > > > > > > > > In the above command, I just add another table reference to it and it gives > me two different results.
I'm not 100% sure what you mean, but if you mean that basically adding the reference to test.country to a statement that looked like "select A.* from test.currency A where A.curr_cd='USD'", yes it gives different results, but it also means something different. > Even I add columns like B.* to it, it do the same things too. Is it not > consistance? > In real world, I don't know the 'country' table is empty or not. Well, you have to write your queries to do what you want depending on such things. For example, the above doesn't constrain the join from currency and country and so you get multiple copies of the USD currency info for each country. If you want to constrain the currency and country information (for example, say A.country=B.id if you had that sort of information) you need to decide what happens if there is no country that matches the country. ---------------------------(end of broadcast)--------------------------- TIP 6: explain analyze is your friend
