Question

WHAT I’M DOING:

I’m writing a C# application which will help employees process emails. I have a database with mails (MS SQL Server 2008 Express). One table contains the mails themselves (each row has columns like msgId, sender, recipients, subject, body etc) and the other table contains data of how the mails have been processed by employees – whenever an email has some property assigned, a new row with the property is saved to the second table. The second table contains columns: msgId, forTeam, notForTeam, processedByTeam. Whenever an employee marks a message as related to his team, a row is saved with TEAMXX value in the forTeam column. Whenever an employee performs other actions on the email, a row with TEAMXX value is saved in the processedByTeam column.

I’m trying to get emails which have following values:
forTeam: ‘TEAM01’
notForTeam: everything different than ‘TEAM01’
processedByTeam: everything different than ‘TEAM01’

I do a LEFT JOIN on the Mails table with the MailAssignments table and then after WHERE I put the above conditions. The result is supposed to show messages which have been assigned to TEAM01 but have not been yet processed by the employees from this team.

THE PROBLEM:

I can’t get the SQL query to work properly. Any combination of the conditions will return extra rows (eg. with property notForTeam=’TEAM01′ when I want to omit those!) or give me an error that datetime is in incorrect format for some reason (I’m sorting emails by received time).

Example query:

SELECT TOP 30 Mails.* 
FROM Mails 
LEFT JOIN MailAssignments 
ON Mails.msgId = MailAssignments.msgId 
WHERE
(MailAssignments.forTeam='TEAM01') 
AND (MailAssignments.notForTeam<>'TEAM01' OR MailAssignments.notForTeam IS NULL) 
AND (MailAssignments.processedByTeam<>'TEAM01' OR MailAssignments.processedByTeam IS NULL)
ORDER BY Mails.sortingTime DESC

Even if I reduce the conditions after WHERE (and before ORDER BY) to (MailAssignments.notForTeam<>'TEAM01') for testing, I still get the error for some reason.

What am I missing or doing wrong?

UPDATE:

It seems that (MailAssignments.notForTeam<>'TEAM01') is somehow causing the problem. Everything else works OK (also with the equals sign is OK: MailAssignments.notForTeam=’TEAM01′) but when I add the notForTeam<> condition I get the incorrect datetime error (WTF? :/)

System.Data.SqlTypes.SqlTypeException: SqlDateTime overflow. Must be between 1/1/1753 12:00:00 AM and 12/31/9999 11:59:59 PM.

UPDATE – EXAMPLE:

table with mails ("Mails"):
msgId sender  subject body  received (sortingTime)
53    [e-mail] test    test  2012-05-11 11:00
54    [e-mail] test2   test2 2012-05-10 10:00
55    [e-mail] blah    blah  2012-05-11 12:00
56    [e-mail] dfgg    dfgg  2012-05-11 12:30


table with assignments ("MailAssignments"):
msgId forTeam notForTeam  processedByTeam
54    null    TEAM02      null            - this means TEAM02 hid the mail from their view
54    TEAM01  null        null            - this means TEAM01 assigned the mail to them
54    null    null        TEAM01          - TEAM01 has processed the mail

53    TEAM01  null        null            - invisible to other teams
53    null    TEAM01      null            - invisible for TEAM01 (and also for others because of the above)

56    TEAM01  null        null
56    null    null        TEAM01          - mail processed by TEAM01

55    TEAM02  null        null            - assigned to TEAM02 but not processed yet


Now, I want my C# app to have a view which will display mails as "This Team's", ie.:
1. mail 54 was first hidden for TEAM02 - should be invisible for them
2. mail 54 was assigned to TEAM01 - should be now visible for them in "This Team's" view
3. mail 54 was processed by TEAM01 - should be now invisible in "This Team's" view

4. mail 53 was assigned to TEAM01 - should visible in their "This Team's" view (and ONLY for them)
5. mail 53 was hidden for TEAM01 - should be invisible to TEAM01 (and for anyone else because of prev. action)

6. mail 56 was assigned to TEAM01 - should be now visible only to TEAM01 in "This Team's"
7. mail 56 was processed by TEAM01 - should be now invisible for TEAM01 in "This Team's"

So, with the present data in both tables, finally, TEAM02 for example should see in their "This Team's" view only:
mail 55

Answer 1

Select TOP 30 *
From Mails As M
Where Exists    (
                Select 1
                From MailAssignments As MA1
                Where MA1.msgid = M.msgid
                    And MA1.forTeam = 'TEAM01'
                Except
                Select 1
                From MailAssignments As MA1
                Where MA1.msgid = M.msgid
                    And ( MA1.processedByTeam = 'TEAM01' 
                         Or MA1.notForTeam = 'TEAM01' )
                )
Order By M.received Desc

SQL Fiddle version.

Having the sample data and schema makes all the difference in the world. What was missed is that there are multiple MailAssignments rows for each Mails row. So, when you say that you want to exclude rows where notForTeam = 'TEAM01', what that really means is to exclude any associated msgid that has that value. I’ve updated my query to accommodate that request. The Except clause as you might suspect eliminates any matching values from the second query that exist in the first query. Since I’m using a constant in both queries, if the second query returns any rows for the given msgid, it will make the entire Exists query return no rows and exclude the msg.

That said, we get no rows for TEAM01 because:

MsgId = 54 is excluded because according to Rule #3, messages 
    that were processed by the team should be excluded.
MsgId = 53 is excluded because notForTeam = 'TEAM01' (Rule #5)
MsgId = 56 is excluded because processedBy = 'TEAM01' (Rule #7)
MsgId = 55 is excluded because forTeam = 'TEAM02'