start migration more web endpoints to sqlx

[syphar]

refs #874

This migrates more web endpoints to sqlx.

Notable omissions around Limits and CrateDetails, which should be async too, but I wanted to leave them for another PR.

Also I want to find a better way to use async in tests, bestrefs #874

This migrates more web endpoints to sqlx.

Notable omissions around Limits and CrateDetails, which should be async too, but I wanted to leave them for another PR.

Also I want to find a better way to use async in tests, best case even using tokio::test or even sqlx::test, at the least some async_wrapper that enables you to write .await in tests. Also for a separate PR. case even using tokio::test or even sqlx::test, at the least some async_wrapper that enables you to write .await in tests. Also for a separate PR.

Last think I also didn't do is parallelization of independent queries in a single handler, which could be investigated separately.

[syphar]

interesting test failure, I remember running the tests locally and it worked, will try again fixed

[Nemo157]

at the least some async_wrapper that enables you to write .await in tests.

I had a quick look at this. The biggest blocker is that all the database calls in the test fakes need to be moved over too, otherwise they error like

thread 'web::releases::tests::get_releases_by_stars' panicked at postgres-0.19.7/src/config.rs:449:44:
Cannot start a runtime from within a runtime. This happens because a function (like `block_on`) attempted to block the current thread while the thread is being used to drive asynchronous tasks.

It's also non-trivial because of the higher-ranked lifetimes meaning that closures don't work, I think the best approach might be to use a macro to wrap the tests giving an API like

async_tests! {
    async fn foo(env: &TestEnvironment) -> Result<()> {
        ...
    }
}

You can see my experiment in the wrapper here, with the inner async fn used instead of a closure to have the lifetimes work: master...Nemo157:docs.rs:test-async-wrapper.

[syphar]

It's also non-trivial because of the higher-ranked lifetimes meaning that closures don't work, I think the best approach might be to use a macro to wrap the tests giving an API like

I remember I made it work by using an async closure + Rc for the TestEnvironment, but in general I'll look into this separately.

I had a quick look at this. The biggest blocker is that all the database calls in the test fakes need to be moved over too, otherwise they error like

yep, the sync/async mix makes some things a little ugly until more is migrated.

[syphar]

I accidentally amended other changes from #2312 to the release-time-nullable commit, so now I just squashed them all to avoid more confusion.