Helen Borrie wrote:

> At 11:01 PM 11/11/2003 -0500, you wrote:
>
>
>>>Use RDB$DATABASE instead of ABNTRADEDATA as the last reference - there is
>>>only one record in RDB$DATABASE
>>>
>>>SELECT (select avg(volume) from ABNTRADEDATA where volume >0), (select
>>> avg(volume) from abntradedata where volume <0) AVG_VOLUME from
>>> RDB$DATABASE;
>>>
>>>Alan
>>>
>>
>>I could be Forrest Gumping here,
>
>
> Yes. :-))
>
>
>>but I don't get how RDB$Database
>>relates to the query.... Just sort of a place holder?
>
>
> In fact, RDB$Database is a system table that hold one and only one
> record. When we use it this way, we care not what it stores. Its
> usefulness is *because* it contains one and only one record.
>
> You could create your own one-record table called DUAL, if you liked. We
> use rdb$database the same way Oracle users use DUAL. :-)
>
> As long as you only want to get computed results back, the only important
> thing is to make sure you do the query FROM a single-record table. If you
> use any table with multiple records, you'll get multiple (identical) records.
>
> So it's a useful technique when you want to pull things like subquery
> results, context variable values, generator values, using a SELECT
> statement, without FROMming a meaningful table reference.

Nice explanation. Thanks Helen. I learned much from reading the
Borland Interbase groups, but its cool information like this that makes
this list fun and very informative.

--

Warm Regards,

Lee