Move first query builder instantiation out of init and into Model.boot() #623
Conversation
This allows for logic affecting the query builder (such as db connection resource injection) to be injected by observer's "booting" event
|
hmm interesting. Tests seem to pass 👍 |
|
was there any additional manual testing you did to prove this works? |
|
The docs say that the tests were written without need for a db (https://github.com/MasoniteFramework/orm/blob/2.0/CONTRIBUTING.md#running-tests), but in my experience, it wasn't working without configuring databases. There's probably some step I was missing. I have another branch I'm working on trying to get local tests passing with docker containers: https://github.com/stoutput/orm/tree/fix-local-tests. My branch was working a week or so ago but I made some changes along the way that broke it. It's almost there, and would be nice to not have to rely on a local MySQL instance. |
|
I would have liked to add a test case asserting that the DB config wasn't accessed until after the "booting" event had fired, but otherwise, I wouldn't expect this to break any existing tests. We're lazily instantiating it the first time we try to access it, during the |
|
Agreed. I just wanna take a deeper look at this tonight and make sure theres no hidden assumptions in the architecture that would break this. |
This allows for logic affecting the query builder (such as db connection resource injection) to be injected by observers' "booting" events