I have refactored all schema creation code into a new class.
It helps keeping the other classes more focused and reduces duplication a bit.
👉 I made it public, so that it can be used to create the schema at another time / with another user.
👉 I also moved all DDL statements in a single file with nothing else, so that they are easy to pick up if someone wants to create the schema through another mean (e.g. running scripts manually with an admin user). Security best practices ;)
❌ The scripts (and code for that matter) are full of inconsistencies (e.g. varchar vs nvarchar2, naming, ...) but this PR does not change them.
👉 I changed the way existing tables are detected to make it a lot more efficient in large schemas.
Above changes fix #10.
👉 I added a flag to not create automatically the transport schema. Somehow, it was the only part that did not have such a flag (did I say consistency?).
👉 I added support for 2-part names such as Rebus.Saga.
This is useful if you want to re-use the same Oracle connections as your application code but still keep each app and rebus in their own schema (instead of opening 2 connections with different users).
But the real motivation is in preparation of #13. When you enlist with the app's own logic, you won't have an option to open different connections anymore.
📢 Rebus.Saga works but take notice:
Rebus DB user won't be created automatically (that's nasty in Oracle, it's more than a SQL Server schema). It has to exist ahead of time.
Of course the current user needs appropriate grants to create DB objects in the respective schema (if you let Rebus create them); plus grants to read/write inside them. This library does not grant rights.
❌ Quoted names like Rebus."Saga Snapshot" did not work before, they still don't.
This is done on top of #18
I have refactored all schema creation code into a new class. It helps keeping the other classes more focused and reduces duplication a bit.
👉 I made it public, so that it can be used to create the schema at another time / with another user.
👉 I also moved all DDL statements in a single file with nothing else, so that they are easy to pick up if someone wants to create the schema through another mean (e.g. running scripts manually with an admin user). Security best practices ;)
❌ The scripts (and code for that matter) are full of inconsistencies (e.g.
varcharvsnvarchar2, naming, ...) but this PR does not change them.👉 I changed the way existing tables are detected to make it a lot more efficient in large schemas.
Above changes fix #10.
👉 I added a flag to not create automatically the transport schema. Somehow, it was the only part that did not have such a flag (did I say consistency?).
👉 I added support for 2-part names such as
Rebus.Saga.This is useful if you want to re-use the same Oracle connections as your application code but still keep each app and rebus in their own schema (instead of opening 2 connections with different users). But the real motivation is in preparation of #13. When you enlist with the app's own logic, you won't have an option to open different connections anymore.
📢
Rebus.Sagaworks but take notice:RebusDB user won't be created automatically (that's nasty in Oracle, it's more than a SQL Server schema). It has to exist ahead of time.❌ Quoted names like
Rebus."Saga Snapshot"did not work before, they still don't.✔️ 79/79 tests green