Ive inherited a table called contacts which is used throughout the
site. There are a lot of duplicates in this table we want to
remove(i.e. the email field was not set to unique so multiple accounts
were created etc etc).
The plan is to select the contact we want to keep and then select the
ones we want to remove and transfer the needed data over before
deleting the unwanted records. The trouble with this is that there a
lots of associated tables, complexed self referential joins etc that
the contacts table relates to. If its not done correctly then further
down the line there could be serious issues as data may be lost etc…
A thought of mine was to simply merge them but instead of deleting the
record we would point it to the main record we are keeping and then
set the records we dont need to “inactive” or something so we know
they are no longer used and point to an active record.
I can think of how to set it this up in the table but can anyone think
of a super cool way to make sure the inactive record always returns
the active record it points to when ever it its referenced or needed?
Any advice appreciated,
JB
On Thu, Apr 8, 2010 at 8:20 AM, johnnybutler7
[email protected]wrote:
If its not done correctly then further
down the line there could be serious issues as data may be lost etc…
I don’t think you’re realistically going to avoid having a huge headache
if
you don’t do the migration correctly the first time. And if you try to
hack
in some automatic record reconciliation logic into your system, you will
forever pay the price of that in terms of application complexity,
performance, inflexibility, and unmaintainability moving forward.
My recommendation is to make extra sure your migration logic from the
old
schema/table to the new one is 100% correct, then backup your DB, do
your
migration, and live a happy new life. Get your data clean, don’t hamper
yourself trying to sit on the fence between fixing the data and leaving
it
in a broken state.
Also, I’m not aware of any technology (Rails or DB-level) which will let
you
somehow automatically alias old rows to new rows when doing joins. Joins
are
not that smart. You’d probably need to write custom SQL which gets the
initial results, then checks them for bad rows, then patches in correct
data
where needed, etc. It’d be a mess!
my 2c
jsw