Disabling lazy loading in provide

[IvanKirpichnikov]

Idea:
I suggest adding the ability to disable lazy dependency loading for the 'provide' function or globally for the entire 'Provider'

How is it supposed to work?
When creating a new container, create all dependencies with the 'lazy=False' flag and add them to the cache.

Why do you need it?
Create resource-intensive sessions or pools, as well as detect errors early when creating dependencies and protect against errors during code execution

How I see it in the code
Provide

class MyProvider(Provider)
    a = provide(A, scope=Scope.App, lazy=False)

Provider

class MyProvider(Provider)
    lazy = Fasle
    a = provide(A, scope=Scope.App)

[Tishka17]

Notes:

1. we probably cannot implement it for async container as it requires async loop, which is often not available at creation point
2. many real objects are lazy: sqlalchemy engine does not open connections on creation, api clients do not do any requests. So the logic could be more complicated. It can be moved to factory, but anyway
3. in some cases (lambda functions, pre-fork model in gunicorn and celery with processes) we still want to have lazy initialisation

[Tishka17]

We can add separate method, which creates all dependencies of current scope. As it will be called after creation it can be async

[tuukkamustonen]

Was there such a method nowadays? Something which would initialize all dependencies in a container?

Use case: Pre-load/warm a web application before taking it online.

[Tishka17]

No, there is no additional method because someone would like to specify

• which objects should be warmed
• additional actions to do

the easiest way is to do it "manually" in your framework startup event. Just call container.get() with proper classes and do what you believe useful in your case (like select 1 or other service pings).

as an option - do it in "readyness" probe handler instead of startup

[tuukkamustonen]

which objects should be warmed

I'd just load up it all, why skip something?

I'm sure there are use cases with more granular needs, too. But the basic use case would to initialize everything, no?

In our case, we have like 50+ "main level" dependencies (these being controller classes for the endpoints), and re-declaring those would be somewhat annoying.

Of course, I could define the controllers in a separate list, which I can then feed to the provider and also read in the loop which get()s everything. It would be nicer to have that "load it all up" method, though.

what you believe useful in your case (like select 1 or other service pings).

I'm actually hoping that could be more automated, too, like in #198 (I'll add a comment there, too).

do it in "readyness" probe handler instead of startup

What would that solve?

I don't think there's benefit in get()'ing everything in readiness probe - all of that would just come from cache.

It should be sufficient to initialize things in the startup probe, and after that the container is already warm so getting they all up again each time would not give anything (except if that triggered pings/health checks).

[tuukkamustonen]

I took a look at #114 and did something like:

        for key in container.registry.factories:
            ...potential filtering...
            await self.adi._get(key)  # noqa: SLF001

Seems functional and quite simple.

[Tishka17]

@tuukkamustonen do not forget objects with another scopes