Обсуждение: Do I just not understand count()?
If I have the table t defined as: a --- 1 1 2 and I say: select count(a=1) from t; should it give me 1 or 2 as a result? I'm getting 2, and I'd think I should get 1....
I don't think I've seen that particular syntax used before (I would say select count(a) from t where a=1;), but since the query appears to work, I won't argue. Why do you think it should give you a result of 1? There are two rows containing a value of 1 for a, hence it returns 2. Greg ----- Original Message ----- From: "Ben" <bench@silentmedia.com> To: <pgsql-general@postgresql.org> Sent: Monday, April 08, 2002 2:25 PM Subject: [GENERAL] Do I just not understand count()? > If I have the table t defined as: > > a > --- > 1 > 1 > 2 > > > and I say: > > select count(a=1) from t; > > should it give me 1 or 2 as a result? I'm getting 2, and I'd think I > should get 1.... > > > ---------------------------(end of broadcast)--------------------------- > TIP 4: Don't 'kill -9' the postmaster >
Er, uh, well, because I confused myself with my trivial example. :) What I should have said was: it's giving me 3 (not 2), and I'm expecting 2 (not 1). On Mon, 8 Apr 2002, Hunter Hillegas wrote: > Why would you think it would return 1? the number of rows where a=1 in t is > 2... > > > > From: Ben <bench@silentmedia.com> > > Date: Mon, 8 Apr 2002 11:25:43 -0700 (PDT) > > To: <pgsql-general@postgresql.org> > > Subject: [GENERAL] Do I just not understand count()? > > > > If I have the table t defined as: > > > > a > > --- > > 1 > > 1 > > 2 > > > > > > and I say: > > > > select count(a=1) from t; > > > > should it give me 1 or 2 as a result? I'm getting 2, and I'd think I > > should get 1.... >
That's what I'd normally do to, but in this case I want to run a query more like select count(a=1), count(a=2) from t and I don't want to do multiple selects, because I'm selecting other stuff too, which takes time, and I figure as long as postgres is looking at those rows, it might as well tally up the counts of a=1 and a=2. On Mon, 8 Apr 2002, Gregory Wood wrote: > I don't think I've seen that particular syntax used before (I would say > select count(a) from t where a=1;), but since the query appears to work, I > won't argue. > > Why do you think it should give you a result of 1? There are two rows > containing a value of 1 for a, hence it returns 2. > > Greg > > ----- Original Message ----- > From: "Ben" <bench@silentmedia.com> > To: <pgsql-general@postgresql.org> > Sent: Monday, April 08, 2002 2:25 PM > Subject: [GENERAL] Do I just not understand count()? > > > > If I have the table t defined as: > > > > a > > --- > > 1 > > 1 > > 2 > > > > > > and I say: > > > > select count(a=1) from t; > > > > should it give me 1 or 2 as a result? I'm getting 2, and I'd think I > > should get 1.... > > > > > > ---------------------------(end of broadcast)--------------------------- > > TIP 4: Don't 'kill -9' the postmaster > > >
> If I have the table t defined as: > > a > --- > 1 > 1 > 2 > > > and I say: > > select count(a=1) from t; > > should it give me 1 or 2 as a result? I'm getting 2, and I'd think I > should get 1.... It will give you a result of 3, as a=1 does not toss out rows for which a <> 1. You would need to add some parameters to limit the rows that count is countin... select count(a) from t where a = 1 would give you 2 select count(DISTINCT a) from t where a = 1 would give you 1 > > > ---------------------------(end of broadcast)--------------------------- > TIP 4: Don't 'kill -9' the postmaster
I'm fairly sure you can't return a conditional count... count likes to merely return the number of rows that match the query, which in this case is all of them. I'm not sure exactly what it does with "a=1" though... I'm guessing that it returns a boolean value, which means that it counts as a row (only NULL values are not counted in count()). If you're selecting other stuff, you won't be able to return more than one row anyway... you'll have to either use a subquery or do a GROUP BY. Greg ----- Original Message ----- From: "Ben" <bench@silentmedia.com> To: "Gregory Wood" <gregw@com-stock.com> Cc: "PostgreSQL-General" <pgsql-general@postgresql.org> Sent: Monday, April 08, 2002 2:45 PM Subject: Re: [GENERAL] Do I just not understand count()? > That's what I'd normally do to, but in this case I want to run a query > more like > > select count(a=1), count(a=2) from t > > and I don't want to do multiple selects, because I'm selecting other stuff > too, which takes time, and I figure as long as postgres is looking at > those rows, it might as well tally up the counts of a=1 and a=2. > > On Mon, 8 Apr 2002, Gregory Wood wrote: > > > I don't think I've seen that particular syntax used before (I would say > > select count(a) from t where a=1;), but since the query appears to work, I > > won't argue. > > > > Why do you think it should give you a result of 1? There are two rows > > containing a value of 1 for a, hence it returns 2. > > > > Greg > > > > ----- Original Message ----- > > From: "Ben" <bench@silentmedia.com> > > To: <pgsql-general@postgresql.org> > > Sent: Monday, April 08, 2002 2:25 PM > > Subject: [GENERAL] Do I just not understand count()? > > > > > > > If I have the table t defined as: > > > > > > a > > > --- > > > 1 > > > 1 > > > 2 > > > > > > > > > and I say: > > > > > > select count(a=1) from t; > > > > > > should it give me 1 or 2 as a result? I'm getting 2, and I'd think I > > > should get 1.... > > > > > > > > > ---------------------------(end of broadcast)--------------------------- > > > TIP 4: Don't 'kill -9' the postmaster > > > > > > >
Try select sum(case when a=1 then 1 else 0 end), sum(case when a=2 the 1 else 0 end) from t; to get the number of occurrences of each value. Jim Ballard ----- Original Message ----- From: "Clinton Adams" <clinton@vote-smart.org> To: "Ben" <bench@silentmedia.com> Cc: <pgsql-general@postgresql.org> Sent: Monday, April 08, 2002 12:49 PM Subject: Re: [GENERAL] Do I just not understand count() > > If I have the table t defined as: > > > > a > > --- > > 1 > > 1 > > 2 > > > > > > and I say: > > > > select count(a=1) from t; > > > > should it give me 1 or 2 as a result? I'm getting 2, and I'd think I > > should get 1.... > It will give you a result of 3, as a=1 does not toss out rows for which a <> > 1. > > You would need to add some parameters to limit the rows that count is > countin... > select count(a) from t where a = 1 > would give you 2 > > select count(DISTINCT a) from t where a = 1 > would give you 1 > > > > > > > > > > ---------------------------(end of broadcast)--------------------------- > > TIP 4: Don't 'kill -9' the postmaster > > ---------------------------(end of broadcast)--------------------------- > TIP 2: you can get off all lists at once with the unregister command > (send "unregister YourEmailAddressHere" to majordomo@postgresql.org) >
Perhaps it would clear up things for those more expert on SQL [and it's variations across DBs] than I if you could say on what system you would normally do this. > > That's what I'd normally do to, but in this case I want to run a query > > more like > > > > select count(a=1), count(a=2) from t > > > > > > If I have the table t defined as: > > > > > > > > a > > > > --- > > > > 1 > > > > 1 > > > > 2 > > > > > > > > > > > > and I say: > > > > > > > > select count(a=1) from t; > > > > > > > > should it give me 1 or 2 as a result? I'm getting 2, and I'd think I > > > > should get 1.... > > > > -- Nigel J. Andrews Director --- Logictree Systems Limited Computer Consultants
"Gregory Wood" <gregw@com-stock.com> writes:
> I'm not sure exactly what it does with "a=1" though... I'm
> guessing that it returns a boolean value, which means that it counts as a
> row (only NULL values are not counted in count()).
Yup, Greg gets a gold star: COUNT *only* cares whether its input is NULL
or not, not what specific value it might have. This is per SQL spec.
The nearby suggestions involving SUM() look like they would work to
accumulate counts of different conditions in a single pass.
regards, tom lane
This is the first time I've tried. I noticed that I could add a DISTINCT clause to count(), and wondered why I couldn't add an equality and make it work as well. On Mon, 8 Apr 2002, Nigel J. Andrews wrote: > > > Perhaps it would clear up things for those more expert on SQL [and it's > variations across DBs] than I if you could say on what system you would > normally do this. > > > > > That's what I'd normally do to, but in this case I want to run a query > > > more like > > > > > > select count(a=1), count(a=2) from t > > > > > > > > If I have the table t defined as: > > > > > > > > > > a > > > > > --- > > > > > 1 > > > > > 1 > > > > > 2 > > > > > > > > > > > > > > > and I say: > > > > > > > > > > select count(a=1) from t; > > > > > > > > > > should it give me 1 or 2 as a result? I'm getting 2, and I'd think I > > > > > should get 1.... > > > > > > > > > -- > Nigel J. Andrews > Director > > --- > Logictree Systems Limited > Computer Consultants > > > > ---------------------------(end of broadcast)--------------------------- > TIP 5: Have you checked our extensive FAQ? > > http://www.postgresql.org/users-lounge/docs/faq.html >