Triggers and Flag bit

Thread view:

Enable EMail Alerts

Start New Thread

Thread rating:

Sharif Islam - 24 Mar 2008 22:14 GMT

I have a perl script (uses sql query) that will run if there is any
update in the row. Currently, I am a trigger that sets the DateModified
field with getdate() Then my perl script looks for certain range of
modified dates and runs the necessary query.

I am trying to come up with a different mechanism where I don't have to
use the date field. I am looking into flag bits. But not sure how
to use it.

Should I create a trigger that will set the flag bit after any update?
This way the script will just look for the updated records, regardless
of what time it was updated. Maybe I am misunderstanding flag bits, then
after another subsequent update how would the perl/sql script know which
records were updated? I hope this make sense.

thanks.

--s

Reply to this Message

Erland Sommarskog - 24 Mar 2008 22:53 GMT

> I have a perl script (uses sql query) that will run if there is any
> update in the row. Currently, I am a trigger that sets the DateModified
[quoted text clipped - 10 lines]
> after another subsequent update how would the perl/sql script know which
> records were updated? I hope this make sense.

There are a couple of alternatives. Which version of SQL Server are you
using? If you use SQL 2005, Query Notification is an interesting
alternative. Your Perl script would issue a query, and then wait until
the result set changes. Of course, this presumes that you use an API
that supports Query Notification, but Win32::SqlServer, available from
my web site on http://www.sommarskog.se/mssqlperl/index.html does.

Another alternative is to use a timestamp column. A timestamp column
in SQL Server has nothing do with date and time, but is automatically
updated each time a row is updated with a database-unique value that
grows monotonically. Thus, the Perl script could look at the timestamp
column and save the latest value as a high-water mark. This would at
least save you the trigger.

I don't believe in flag bits. In this case, the Perl script would have
to flip them back, and that's more complex.

Signature

Erland Sommarskog, SQL Server MVP, esquel@sommarskog.se

Books Online for SQL Server 2005 at
http://www.microsoft.com/technet/prodtechnol/sql/2005/downloads/books.mspx
Books Online for SQL Server 2000 at
http://www.microsoft.com/sql/prodinfo/previousversions/books.mspx

Reply to this Message

Sharif Islam - 25 Mar 2008 17:22 GMT

>> I have a perl script (uses sql query) that will run if there is any
>> update in the row. Currently, I am a trigger that sets the DateModified
[quoted text clipped - 10 lines]
>> after another subsequent update how would the perl/sql script know which
>> records were updated? I hope this make sense.

[...]

> Another alternative is to use a timestamp column. A timestamp column
> in SQL Server has nothing do with date and time, but is automatically
> updated each time a row is updated with a database-unique value that
> grows monotonically. Thus, the Perl script could look at the timestamp
> column and save the latest value as a high-water mark. This would at
> least save you the trigger.

Thanks, the timestamp solution seems feasible. I am little confused on
how to use it.

According to the documentation SELECT @@DBTS should return the records
that were just updated or modified. After some updated, when I ran the
select statement, I am getting all the rows. Is there anything else I
need to do? When I created the field I just assigned the data type,
didn't include any default vaule (using MSSQL 2005).

--s

Reply to this Message

Erland Sommarskog - 25 Mar 2008 23:35 GMT

> Thanks, the timestamp solution seems feasible. I am little confused on
> how to use it.
[quoted text clipped - 4 lines]
> need to do? When I created the field I just assigned the data type,
> didn't include any default vaule (using MSSQL 2005).

Not sure what you mean here, as SELECT @@DBTS returns a single value.
Probably you ran some other query, but as I don't see it, I can't say
what's wrong with it. :-)

Rather than using @@DBTS, min_active_rowversion() is a better choice.
This function was added in SP2, and it was added to Books Online as late
in the September 2007 edition (see my signature for download link). Looking
at @@dbts can cause some issues when there are uncommitted transaction.

Signature

Reply to this Message

Sharif Islam - 28 Mar 2008 16:44 GMT

>> Thanks, the timestamp solution seems feasible. I am little confused on
>> how to use it.

> Rather than using @@DBTS, min_active_rowversion() is a better choice.
> This function was added in SP2, and it was added to Books Online as late
> in the September 2007 edition (see my signature for download link). Looking
> at @@dbts can cause some issues when there are uncommitted transaction.

Thanks for the help. Here's how I am using it. I created a column 'Flag'
with timestamp datatype.

------

declare @before timestamp
declare @after timestamp
set @before= min_active_rowversion() -1
update MyTable set MyCol ='Test' where MyCol like 'Test123%'
set @after = min_active_rowversion() -1

select ID,MyCol from MyCol where Flag -1 < @after
and Flag -1 >= @before
-----
This gave me the list of record ID that was just changed. Is this the
way to use min_active_rowversion()?

--s

Reply to this Message

Erland Sommarskog - 28 Mar 2008 23:04 GMT

> Thanks for the help. Here's how I am using it. I created a column 'Flag'
> with timestamp datatype.
[quoted text clipped - 12 lines]
> This gave me the list of record ID that was just changed. Is this the
> way to use min_active_rowversion()?

Hm, not really.

Your Perl script would run a batch like:

DECLARE @new_highwater_mark rowversion
SELECT @new_highwater_mark = min_active_rowversion()

SELECT ID, MyCol, @new_highwater_mark
FROM tbl
WHERE tstamp >= @last_highwater_mark AND
tstamp < @new_highwater_mark

That is, the Perl script needs to remember @new_highwater_mark, and pass
it as @last_highwater_mark next time.

It's important to capture min_active_rowversion() into a variable, because
it could change while the query is running, which could lead to lost
updates.

Signature

Reply to this Message

Sharif Islam - 31 Mar 2008 02:19 GMT

>> Thanks for the help. Here's how I am using it. I created a column 'Flag'
>> with timestamp datatype.
[quoted text clipped - 31 lines]
> it could change while the query is running, which could lead to lost
> updates.

Got it! Thanks a lot.

Reply to this Message

--CELKO-- - 25 Mar 2008 21:09 GMT

>> I am trying to come up with a different mechanism where I don't have to use the date field [sic: columns are not fields]. I am looking into flag bits. <<

All that will do is destroy data you have collected. Just use the
date.

Reply to this Message