Activity overview
Latest activity by mark__a
james.billings wrote:
Unfortunately if the foreign keys are there, then it'll always use the FK generator. If you're copying records across for the other two tables does this matter though? Presumably, the primaries in the other two tables are what you want anyway?
This is a serious limitation in your product. All I really want to do is to "UPDATE" a few columns to wipe out some of the production data columns so that we can have a realistic test database.
As a concrete example for why this is a serious limitation, I put forth the following:
Customer can have 0 or more related Address records (via CustomerAddresses). If the CustomerAddresses CustomerId and/or AddressId FKs are randomly assigned, then the Addresses won't make much sense at all. You'll have Address records from different states or countries. Which is fine if you're trying to test for all available inputs but terrible if your goal is a realistic-yet-cleansed set of data.
There are a number of other examples in our database where many-to-many relationships only make sense if you keep the same keys. As far as I can tell, the only way to do this is to run a post-SQL Generator script that deletes all the FKs in these m2n tables, sets IDENTITY INSERT and then re-imports the existing data.
Why can't we just import all of the CustomerAddresses FKs as-is? Can't we just set the generation order so that Customer and Address are generated before CustomerAddresses? / comments
james.billings wrote:
Unfortunately if the foreign keys are there, then it'll always use the FK generator. If you're copying records across for the other two tables does this matter though? Pres...
BenDippenaar wrote:
It would be really nice if the data generator could generate data that matches the referential ‘footprint’ of the data it is generating from.
for example, it doesn't seem like i can preserve existing many-to-one relationships in the anonymized database.
We are trying to do exactly the same thing. We have three tables:
Address, Customer, CustomerAddresses (contains only 2 FKs: AddressId and CustomerId). A Customer can have 0 to many Addresses.
The goal is to generate new address info and customer names but keep the same FKs in the CustomerAddresses table. I am using the "Use existing data source" option and am referencing a copy of the database to generate/import the data from Address and Customer. What I want to do is to import the CustomerAddresses table EXACTLY as it is. Same Id values in each row.
In the "real" database, the data looks like this:
Address/AddressId
1 (and then street1, city, state, zip, etc.)
2
3
...
Customer/CustomerId
1 (and then name, credit limit, etc.)
2
3
CustomerAddresses
CustomerId / AddressId
1 1
1 2
2 3
2 4
3 5
3 6
But when we generate the CustomerAddresses table with "Use existing data source" it forces us to use the Foreign Key Generator for these two columns. We can't change it. How can we just use the existing data and not generate anything?
We cannot de-select the CustomerAddresses table for generation because it chokes when trying to delete all the rows from Customer. / comments
BenDippenaar wrote:
It would be really nice if the data generator could generate data that matches the referential ‘footprint’ of the data it is generating from.
for example, it doesn't seem...