Glenn,
First, I think that you would find your own queries easier to debug if you usedsome indenting and line breaks to
organizethe text. Whenever I run into aquery problem, that's what I do and frequently the missed clause or
problemaggregatebecomes obvious.
> Query 1: returns correct number of rows
> select * from overperms where username='test' union select * from
> defaultperms where username='test' and vimname not in (select vimname from
> overperms where username='test');
>
> Query 2: return incorrect number of rows
> create view perms as select * from overperms union select * from defaultperms
> where vimname not in (select vimname from overperms);
>
> select * from perms where username='test';
These two are NOT the same query, so it's unsurprising that the counts come outdifferent. In the second query, you are
excludingALL rows present inoverperms from the count of defaultperms, not just those rows with a usernameof 'test'.
Thus,if vimname <-> username parings are variable, you willindeed get different counts for the first query than the
second.
Also, I don't think that your query structure is optimal. I think you'remaking this harder than it needs to be.
However,I can't tell without seeingyour data structure.
-Josh Berkus
______AGLIO DATABASE SOLUTIONS___________________________ Josh Berkus Complete
informationtechnology josh@agliodbs.com and data management solutions (415) 565-7293 for law firms, small
businesses fax 621-2533 and non-profit organizations. San Francisco