feat: migrations #4
Conversation
bunnet/migrations/runner.py
Outdated
| logger.info("Building migration list") | ||
| names = [] | ||
| for modulepath in path.glob("*.py"): | ||
| names.append(modulepath.name) |
There was a problem hiding this comment.
Hi @roman-right, could __init__.py be excluded here?
The problem we're facing is that only the initial scripts are run (and any later added migration files are ignored in further runs).
This is because the first migration creates following records in migrations_log (_id and ts omitted):
name: 20221103203943_migration_test.py, is_current: false
name: 20221103204129_migration_test_2.py, is_current: false
name: __init__.py, is_current: true
and newly added migration files (which names start with a timestamp) never appear after __init__.py when sorted alphabetically.
We wish to have __init__.py in the package along with the migration files in order to be able to import the migration files in unit tests.
There was a problem hiding this comment.
Hi @PetrF0X ,
Good point! I'll exclude init file there. Thank you.
| To create a new migration, run: | ||
|
|
||
| ```shell | ||
| bunnet new-migration -n migration_name -p relative/path/to/migrations/directory/ |
There was a problem hiding this comment.
Hey @roman-right,
I couldn't find the code that handles this command so I'll stick my comment here - we've noticed that the generated file name makes it hard for the migration file to be imported in a unit test (the filename starts with a number).
Would you consider a different naming convention for the migration files?
There was a problem hiding this comment.
hm. Yes, you are right. I think, I'll add a prefix like migration_
|
Hi @PetrF0X, |
|
Hi @roman-right, |
No description provided.