Ø Alternatively, the default row should be referenced somewhere else. If this
is per group then you would have a group header table which contains the pk to
the default row in your other table.
This is absolutely correct.
Use another table (with a single row, if that’s what it is) to contain the
‘latest’ or ‘default’ value.
Putting an indicator in the table, which can be ‘Y’ on one and only one row,
will not lead to joy.
Mike
From: oracle-l-bounce@xxxxxxxxxxxxx [mailto:oracle-l-bounce@xxxxxxxxxxxxx] On ;
Behalf Of Andy Sayer
Sent: Thursday, January 30, 2020 7:57 PM
To: Ethan Post <post.ethan@xxxxxxxxx>
Cc: oracle-l <oracle-l@xxxxxxxxxxxxx>
Subject: Re: Update col to Y set all other to N in trigger.
I wouldn’t trust that solution to work well enough when this gets used by more
than one user in a few minutes.
You should code your form so that it submits the update for all necessary rows.
You should also create a unique function based index to enforce this uniqueness
rule. Only 1 row can be the default so you could have the index on
decode(is_default,'Y',1), or if it depends on a group then you would put the
group reference there instead of the hard coded 1.
Alternatively, the default row should be referenced somewhere else. If this is
per group then you would have a group header table which contains the pk to the
default row in your other table.
Thanks,
Andrew
On Fri, 31 Jan 2020 at 00:43, Ethan Post
<post.ethan@xxxxxxxxx<mailto:post.ethan@xxxxxxxxx>> wrote:
In this scenario there is a form user is filling out. When they select the
single row which is default, all other rows must now be not default. I know one
trick (schedule one time job to run right away which sets the other rows to N)
but thought there was a more elegant solution.
On Thu, Jan 30, 2020 at 6:14 PM Andy Sayer
<andysayer@xxxxxxxxx<mailto:andysayer@xxxxxxxxx>> wrote:
Instead of firing an update statement from within the trigger you would just
set the value within the changed row if necessary.
Something like
If :new.pk<http://new.pk> = 1 then
:new.col='Y';
Else
:new.col='N';
End if;
Of course, the other option is to ensure your code to update the table always
goes through some procedure which knows the rules, a so called Transactional
API (XAPI). You could also add a check constraint to enforce the rule so that
an error occurs when an attempt to violate it is made.
Hope that helps,
Andrew
On Thu, 30 Jan 2020 at 23:52, Ethan Post
<post.ethan@xxxxxxxxx<mailto:post.ethan@xxxxxxxxx>> wrote:
I should recall how to do this but been a while.
I need the best way to
update foo set col='Y' where pk=1;
And have a trigger automatically...
update foo set col='N' where pk !=1;
While avoid mutating table error.
This one seems hard to google the answer to.
Thanks,
Ethan
e-t-h-a-n.com<http://e-t-h-a-n.com>
@poststop
