Wrap event::Cucumber in Event to allow metadata

[ilslv]

Synopsis

Despite there is no use for it now, it can become useful later. For example in implementation of JUnit Writer.

Solution

Pair DateTime with every Cucumber event to avoid breaking API in the future.

Checklist

• Created PR:
  • In draft mode
  • Name contains Draft: prefix
  • Name contains issue reference
  • Has assignee
• Documentation is updated (if required)
• Tests are updated (if required)
• Changes conform code style
• CHANGELOG entry is added (if required)
• FCM (final commit message) is posted
  • and approved
• Review is completed and changes are approved
• Before merge:
  • Milestone is set
  • PR's name and description are correct and up-to-date
  • Draft: prefix is removed
  • All temporary labels are removed

This PR is nominated at RU RustCon Contest

[ilslv]

FCM

Wrap `event::Cucumber` into `Event` to support optional metadata (#145)

- add `timestamps` Cargo feature injecting timestamps into `Event`s

[tyranron]

@ilslv I think we can go with a better design. Ping me tomorrow to discuss this in a talk.

[tyranron]

IMHO, this is not very good decision, now we have the code base polluted with those tuples and signatures became worse.

I see here two ways:

1. Attach DateTime directly to events types and errors.
2. Instead of (T, DateTime) pairs have DateTimed<T> wrapper, which encapsulates the stuff and provides obvious interface to inner values.

With the option 2, we will need to only change type signatures, without touching any code. Furthermore, this way we can hide it completely behind a feature gate, so it can be used only for those Writers which really need it.

[ilslv]

@tyranron

Attach DateTime directly to events types and errors.

As all events are nested enums, we'll have to put DateTime in every variant of the Cucumber event

pub enum Cucumber<World> {
    /// [`Cucumber`] execution being started.
    Started(DateTime<Utc>),

    /// [`Feature`] event.
    Feature(DateTime<Utc>, Arc<gherkin::Feature>, Feature<World>),

    /// [`Cucumber`] execution being finished.
    Finished(DateTime<Utc>),
}

Matching over it will be very painful.

Instead of (T, DateTime) pairs have DateTimed wrapper, which encapsulates the stuff and provides obvious interface to inner values.

👍

With the option 2, we will need to only change type signatures, without touching any code.

Not really, especially in case of writer::Normalized. There we store only event::Step and path to it: Feature, Option<Rule>, Scenario. Then we rebuild it with Cucumber::scenario(feat, rule, sc, ev) when it's time to emit it. But it's pretty easy to store DateTimes of those events and already done in this PR.

Furthermore, this way we can hide it completely behind a feature gate, so it can be used only for those Writers which really need it.

I'm not sure we should. This will cause Writer trait to be different in cases when feature is on or off (same for Runner)

pub trait Writer<World> {
    type Cli: StructOptInternal;

    async fn handle_event(
        &mut self,
        ev: parser::Result<event::Cucumber<World>>,
        #[cfg(feature = "event_datetime")] // I don't think it's pleasant API to use
        at: DateTime<Utc>,
        cli: &Self::Cli,
    );
}

[tyranron]

A little bit of bikesheddings, but generally, that's it!

@ilslv good job 👍