Question

I’ve come across a strange query embedded in an application (yeah, great!):

SELECT PersonId
    , Salutation
    , Email
    , Postcode
FROM    vw_NewsletterSubscriptions
WHERE    PersonId IN (SELECT PersonId FROM vw_NewsletterSubscriptions) 
    AND NewsletterTypeID=1 
    AND UnSubscribeDate Is NULL
GROUP BY PersonId
    , Salutation
    , Email
    , Postcode

It’s the SELECT in the WHERE clause that’s got me. It seemed to be saying “select some data from this view where the data is in the view”, which is a bit unnecessary. So I commented out that line in the WHERE clause:

SELECT PersonId
    , Salutation
    , Email
    , Postcode
FROM    vw_NewsletterSubscriptions
WHERE    NewsletterTypeID=1 
    AND UnSubscribeDate Is NULL
GROUP BY PersonId
    , Salutation
    , Email
    , Postcode

For completeness I ran both versions to check they are identical and they are not. The new version returns more rows – rows that were not in the first version. I’ve outer joined the two sets to see the difference but there’s nothing obviously different about the extra rows.

I’m clearly missing something here. Can you explain what’s going on, please?

Answer 1

Thanks for all the ideas, guys. I compared the results of the query with and without the seemingly unnecessary clause and I’m damned if I could see a pattern. The view was an overcomplicated mess so I refactored it and now I’m getting the same number of rows either way. Plus our number of newsletter subscribers just went up too!