Pydantic v2 support

[ManiMozaffar]

Update: Rocketry seems to not be maintained anymore.
If you're looking for an alternative with supporting pydantic v2, try out AioClock.
https://github.com/ManiMozaffar/aioclock

---

Is your feature request related to a problem? Please describe.
The problem is that you're using red-bird for logging which is large code base, and also rocketry is dependent on v1.

Describe the solution you'd like

1. Either create new version of rocketry that matches both pydantic versions, like fastapi did (some dirty trick, but works)
2. Refactor red bird and rocketry to pin pydantic to v2

Describe alternatives you've considered
None

Additional context
None

[mafrosis]

Alternative, and quickest fix:

3. pin Pydantic dependency to 1.10.10

[ManiMozaffar]

Alternative, and quickest fix:

3. pin Pydantic dependency to 1.10.10

Wouldn't work in my case,
Anyway red-bird isn't even maintained by anyone, doesn't seem very much 'alive' library.
What do you think? @mafrosis

[abuchnick-aiola]

Same problem here, this issue is blocking me heavily since June 30th.
If it's relevant and could help, I can submit a PR to add the pydantic v2 support to red-bird
@Miksus

[ManiMozaffar]

Same problem here, this issue is blocking me heavily since June 30th. If it's relevant and could help, I can submit a PR to add the pydantic v2 support to red-bird @Miksus

I didn't notice that red-bird is also maintained by maintainer of this repo,
that's a good news then.
I'll open a PR there, working on Pydantic V2, so hopefully it'll be merged, if not then it should be enough for my own use case at least, to fork rocketry and go with a forked version that supports v2 which uses my own red-bird that supports v2.

[mafrosis]

I think it's up to @Miksus to decide if red-bird is "alive" or not

Simply pin your pydantic version with pydantic==1.10.10. Fixed.

[broven]

Seems I have the similar issue when try intergrate to django followed https://rocketry.readthedocs.io/en/stable/cookbook/django.html

  File "/Users/meta/gitRepos/github.com/broven/ASRHub/.venv/lib/python3.11/site-packages/pydantic/fields.py", line 715, in Field
    raise PydanticUserError('`const` is removed, use `Literal` instead', code='removed-kwargs')
pydantic.errors.PydanticUserError: `const` is removed, use `Literal` instead

For further information visit https://errors.pydantic.dev/2.0.2/u/removed-kwargs

[FrancoTampieri]

I have the same problem, the fix is ok, but for my project I loose all the benefit of pedantic 2.x

[Jypear]

I've started a PR to begin the work integrating pydantic V2 #212
Happy for anybody to contribute with it, I think its going to require a lot of work.

Running locally with a basic 'hello_world' task at the moment, with pydantic 2.0.3 and I've installed @ManiMozaffar red-bird branch where .v1 has been appended to the pydantic imports for now. With Rocketry's dependence on that module, I think that is going to need migrating as well for this to work properly.

[ManiMozaffar]

I've started a PR to begin the work integrating pydantic V2 #212 Happy for anybody to contribute with it, I think its going to require a lot of work.

Running locally with a basic 'hello_world' task at the moment, with pydantic 2.0.3 and I've installed @ManiMozaffar red-bird branch where .v1 has been appended to the pydantic imports for now. With Rocketry's dependence on that module, I think that is going to need migrating as well for this to work properly.

Heads up: use pydantic migration module, so it’ll replace v1 codes with v2. Then 70 tests will fail and you need to start debugging them. Lots works as you mentioned, that’s why i decided to use it with .v1 because internally it doesn’t use pydantic alot, just as passing by arguments to tasks and etc so I didn’t see a great performance boost.

[Jypear]

I've started a PR to begin the work integrating pydantic V2 #212 Happy for anybody to contribute with it, I think its going to require a lot of work.
Running locally with a basic 'hello_world' task at the moment, with pydantic 2.0.3 and I've installed @ManiMozaffar red-bird branch where .v1 has been appended to the pydantic imports for now. With Rocketry's dependence on that module, I think that is going to need migrating as well for this to work properly.

Heads up: use pydantic migration module, so it’ll replace v1 codes with v2. Then 70 tests will fail and you need to start debugging them. Lots works as you mentioned, that’s why i decided to use it with .v1 because internally it doesn’t use pydantic alot, just as passing by arguments to tasks and etc so I didn’t see a great performance boost.

Thanks for the heads up.
Bump script was ran today but it didn't seem to do a whole much. Still working on it, tests are running now atleast.

[Miksus]

Sorry for taking eternity to answer: I should upgrade my machine as developing Rocketry on it is painfully slow with it and have had too busy free time. I'll try to visit this in the week end (and hopefully purchase a new machine).

[Jypear]

Thanks @Miksus
I've been working on this the past couple of nights and seem to have it working (for the most part) on v2 apart from one weird bug (probably not a bug, bug).

The way pydantic has changed has caused some odd class inheritance issues, One of the first things I've had to do is comment out creating the null session in the RedBase class in _base.py (Not an ideal solution)
If we don't then we get a clash issue:

Exception has occurred: NameError
Field name "session" shadows an attribute in parent "RedBase"; you might want to use a different field name with "alias='session'".

This issue was also appearing with other classes further down but I managed to get around it by typing those variables to ClassVar based on a issue I found on the pydantic repo pydantic/pydantic#2410 (comment)

This has been fine for getting the application running but when it comes to the test, because it uses DummyTask which inherits Task/FuncTask and then session is passed into it. This seemed to raise the exact same error as above.

I've tried to get around this by setting ClassVar in the dummy class:

class DummyTask(BaseTask):

    session: ClassVar
    def execute(self, *args, **kwargs):
        return

This then causes another error, where when it then tries to run the validation (on name inside the parse_name function) it doesn't seem to pass this attribute up the inheritance into the validation function. So you just end up getting a key error on the test.

FAILED rocketry/test/condition/test_meta.py::test_taskcond_true[main] - KeyError: 'session'

    @field_validator('name', mode="before")
    def parse_name(cls, value, values):
>       session = values.data['session']
E       KeyError: 'session'

# example test that fails

class DummyTask(BaseTask):

    session: ClassVar
    def execute(self, *args, **kwargs):
        return

def test_defaults(session):
    task = DummyTask(name="mytest", session=session)
    assert task.name == "mytest"
    assert isinstance(task.start_cond, AlwaysFalse)
    assert isinstance(task.end_cond, AlwaysFalse)

All in all, I'm not sure what's changed in pydantic to cause that inheritance issue, but if anyone doesn't find a solution it's probably the last main issue with the PR and I can also reimplement the base class into the class structure. It would just need the rest of the warnings working through (But I think on the tests its inheriting a lot of those from RedBird).

[hlafaille]

Just tried to do a quick fork of this (so I can get back to working) and change the pydantic imports to pydantic.v1, and we need to at least update redbird and tiangolo/pydantic-sqlalchemy.