In that class you can call
Jokes.impl[IO], and the implicits will all be in scope at that point. You can construct a client similarly
I may recommend not using the abstract
F[_] approach if you're new to scala and don't have an experienced fp/cats/etc person to assist
instead just hardcode
IO specifically - it can be changed later if you find it useful
@Daenyth Thanks for the prompt response, tried this , i am not sure what am i missing here
mport java.util.concurrent.Executors
import cats._, cats.effect._, cats.implicits._
import cats.effect.{Blocker, ExitCode, IO, IOApp}
import org.http4s.client.blaze.BlazeClientBuilder
import org.http4s.client.{Client, JavaNetClientBuilder}
import cats._, cats.effect._, cats.implicits._
import scala.concurrent.ExecutionContext.global
object ClientInvoker extends IOApp {
override def run(args: List[String]): IO[ExitCode] = {
Jokes.impl[IO]
}
}runneeds to return an IO[ExitCode].
So
Jokes.impl[IO].get would return you an IO[Jokes.Joke]. Now you've got a joke to do something with. But you still need to get to an ExitCode to represent whether the process succeeded or failed.
What do we do with the joke? Maybe we tell it with
println: joke => IO(println(joke)). That's a Joke => IO[Unit]. And if we already have an IO[Joke], we can flatMap it to tell the joke.
Jokes.impl[IO].get.flatMap(joke => IO(println(joke)). Now we've got an IO[Unit], which represents the telling of the joke.
We did a flatMap to tell the joke, because telling the joke is something else that happens in
IO, because println is a side effect.
And if we have an
IO[Unit], and a Unit => ExitCode, we can get to IO[ExitCode] by a simple map:
So,
Jokes.impl[IO].get.flatMap(joke => IO(println(joke)).map(_ => ExitCode.Success). That should compile. (Maybe not, I'm just typing in Gitter. The compiler is the boss!)
From there, there are various things you can do to clean it up. Factor out the
joke => IO(println(joke)) into a tellJoke(joke: Joke): IO[Unit]. Or .map(_ => ExitCode.Success) can be simplified to .as(ExitCode.Success).
But hopefully that rambling helps clarify some of the types, and how to get from what you have to what you need.
Probably the most confusing thing about that is why
println needs to be wrapped in IO. The compiler is fine if you don't, but that's something that's important when you start to use Scala in a pure FP style. The cats.effect.IO docs have some deeper reading on "Why IO?"
Hopefully that helps. And if not, please ask more questions!
@Ross thanks for detailed explanation, really appreciate this.
I feel terribly asking may be very basic things but i am really struggling with FP and Scala coming from Java world.
Tried below but it does not compile
I feel terribly asking may be very basic things but i am really struggling with FP and Scala coming from Java world.
Tried below but it does not compile
override def run(args: List[String]): IO[ExitCode] = {
Jokes.impl[IO].get.flatMap(joke => IO(println(joke)).map(_ => ExitCode.Success)
}
Looks to me like you're not passing a client to your
Jokes.impl, you need something like override def run(args: List[String]): IO[ExitCode] = {
BlazeClientBuilder[IO](global).resource.use { client =>
Jokes.impl[IO](client).get.flatMap(joke => IO(println(joke))).map(_ => ExitCode.Success)
}
If I want to change the log configuration in https://github.com/http4s/http4s.g8/blob/0.21/src/main/g8/src/main/resources/logback.xml but built an artifact using assembly can we provide that resource seperatly or do I have to rebuild the assembly?
@drocsid Configuring Java loggers has been a miserable process of trial and error for me for well over a decade, but I think you can set a system property to override the one in your assembly.
@hectarV The key concept is once you're in
IO, you stay in IO. It's the responsibility of IOApp to finally run the IO that you produce.
If you have
io: IO[A] and f: A => B, then io.map(f)is IO[B].
If you have
io: IO[A] and f: A => IO[B], then io.flatMap(f) is IO[B].
Those are the two most common ways of "chaining". Note that if you map in the second case, you'd end up with
IO[IO[B]], and you need to "flatten" that. Hence, "flatMap".
And if you're doing a lot of flatMaps in a chian (which is common), that's a good case for for-comprehensions.
Apologies if that's review. Not sure how long you've been studying Scala.
@rossabaker also I have another thing going. I'm deploying via native-image....
and then I invoke the other service with something like:
. . .
postRequest = Request[F](method = POST,
uri = uri" . . .").
withHeaders(`Content-Type`(MediaType.application.`json`)).
withEntity[Json](jsonParams)
result <- client.expect[Json](postRequest)
I noticed the BlazeClient constructor has several parameters like
idleTimeout, requestTimeout and connectTimeout....I am not sure how to change those values and if I could, which one should I change
However, it still seems to ignore those settings...I gave values of several minutes and it always answers
Response timed out after 30 seconds or so...
case AuthedRequest(authToken, req @ POST -> Root / "ageOf") =>