hello anyone can help me please

How can I hate end point with every change without make a lot of requests?

Good morning / good afternoon / good evening! I have a question regarding testing endpoints in ktor and the “new” testApplication. Last time I worked with ktor testApplication didn’t exist. I have this test

@Test
fun `when user user exists returns code 490`() : Unit = testApplication {
    val client = createClient {
        install(ContentNegotiation) {
            json()
        }
    }

    RepositoriesProvider.userRepository.add(User("", "username", "", false, 0, 0))

    val response = <http://client.post|client.post>("/signup") {
        contentType(ContentType.Application.Json)
        setBody(UserRequest("username", "password"))
    }
    assertThat(response.status).isEqualTo(HttpStatusCode.Conflict)
}

to test the end point. I define the testApplication and the client to install the ContentNegotiation. But I have two questions here: 1. Do I need to repeat the createClient code in every method? Is there any way I can wrap the testApplication with this initialization and use it in all the tests. 2. The instance of the userRepository object is different in this test than in the code when the endpoint is reached. Am I missing something in the initialization so I can use the same instance? Thanks in advance!

Hey guys, I'm not sure if this is a Ktor issue, a gradle issue or a Kotlin multiplatform issue, however, it's a fullstack web app project generated by IntelliJ. The way to reproduce the issue is as follows: 1. Have IntelliJ Ultimate installed (newest version up until today) 2. Generate a Kotlin multiplatform project (Ktor backend, Kotlin/JS frontend) 3. Run the

build

Gradle task. 4. Watch the folder

build/libs/

. The

.jar

files get generated. Watch their size. 5. Go back to step 3. Doing this a couple of times causes the .jars to be exponentially large for some reason. I don't know why!

I am trying to create a custom deserialization for an enum. I recognise the enum based on an id. But in the JSON this value can be null, then I would like to take a default enum entry. But here the compile fails if I do the following:

override fun deserialize(decoder: Decoder): MyEnum {
        return if (decoder.decodeNotNullMark()) {
            val enumValue = decoder.decodeInt()
            MyEnum.values().firstOrNull {
                it.id == enumValue
            } ?: MyEnum.DEFAULT
        } else {
            MyEnum.DEFAULT
        }
    }

org.jetbrains.kotlin.backend.common.BackendException: Backend Internal error: Exception during file facade code generation

Anybody an idea how I can fix this?

It would be great if the Ktor changelogs were a bit more “human readable”. https://github.com/ktorio/ktor/blob/main/CHANGELOG.md 2.2.0 indicates that it’s not just a patch release. It’s really hard to distinguish, which of the 67 bullet points are relevant and what’s the larger picture of this release is.

anyone else getting this error on startup after upgrading to ktor 2.2.0?

java.lang.ClassNotFoundException: kotlinx.atomicfu.AtomicFU

Hello ! Ktor-client 2.1.2 is not working on iOS 12. Is there an easy way to find the minimum iOS deployment target for each versions of Ktor-client ?

Is there a way to secure the Swagger UI or OpenAPI with an auth?

routing {
    authenticate("auth-basic") {
        this@routing.swaggerUI(path = "/swagger", swaggerFile = "openapi/documentation.yaml") {
            version = "4.15.5"
        }
    }
}

hey guys, why is open api generator available for embedded server type of app but not when configured by hocon, it’s not even available when the caret is over module function too, how do we generate open api for this use case?

Hello, how would a real world example using websocket and ProtoBuf look like? The examples in the documentation are mimicking what would be API endpoints (imo a really misleading example), while a real world application using websockets would look more like this, were each websocket message needs to be serialized to a different class to consume it. Here it's done by deserializing a Json once to get the message type and then again to get a full typed data class: https://github.com/vaibhavgoyal09/TicTacToeServer/blob/master/src/main/kotlin/com/vaibhav/routes/websockets/GameSocketRoute.kt#L75

What is best way to handle null fields in body?

This message was deleted.

Hello all. how do I create a createApplicationPlugin, which runs after the authentication phase?

hey all. has anyone dealt with latency issues running Ktor server on Google Cloud Run? seeing 5-6s server side latencies to serve 404s and server can only handle 1 request at a time. this is using something pretty close to the CIO server hello world example with gradle :installDist in the container and Ktor 2.0.1. no extra IO is performed on requests and I've ruled out startup time as an issue. latency remains 5-6s regardless of resource limits and the same config runs a NodeJS webserver with no issues. issue is not reproducible locally. anything I could be missing?

Hello all 🙂 I saw that in the last update you've added the Request limiter (max queries per XX unit of time), it is super cool, thanks! I was wondering if you will also consider a concurrent limiter. When average latency is required, it is easier to give the limit of a system by the maximum concurrency (based on the little's law of the queuing theory) It could be used very similarly to the request limiter

Is there any EASY way to get the body's object content from Ktor client's MockEngine's requestHistory?

I think that's one of the most basic needs in testing, and having OutgoingContent there isn't really useful at all...

Hey is there any documentation on the mapping from old api pipelines for ktor client like this one for the server side https://ktor.io/docs/custom-plugins-base-api.html#mapping ?

How do I send json in form data from the ktor client and receive it on a ktor server? I need multiple keys (json and json2) with json data in them in the same request.

How to configure websocket logs in ktor 2.2.1? Is there any docs?

Looking for some insight on this issue. In the IDE, my ktor client's requests go through just fine. After building with jpackage and installing on the same development machine, it appears that no request is ever being made by the client. A thread dump seems to show the thread just waiting forever. I'm unsure why this would be working in the IDE but now in the build, but I think it could be related to how ktor is working with the build tools we're using. Has anyone run into something like this before?

When sending a datagram using UDP, is it possible to determine the port it's sent from?

Since the 2.2.0 update, the

HostRouteSelector::evaluate

function uses

context.call.request.origin.serverPort

to determine if a request matches the

port

routing filter. I’m not sure why it doesn’t call

localPort

instead, as that’s (as far as I understand) the port where the request is actually received. We use a ktor application with CIO in k8s behind mapped ports and an ingress, so the port in the

Host

headers is something completely different from what we specify in our routing. Before 2.2, it worked - probably by accident - because CIO used the local information first. Is there a way to make this work with 2.2? Doesn’t seem like an unusual use case, but I’m probably missing something simple here.

https://ktor.io/docs/swagger-ui.html Is there a way to do add headers to openapi automatically?

Is

HttpClientEngineBase

intended to be subclassed? I ask because

suspend fun execute(data: HttpRequestData): HttpResponseData

is maked as

@InternalAPI

, but

HttpClientEngineBase

itself isn’t.

For newbies to Ktor check out this repo for a simple user authentication using Ktor, FreeMarker and Postgres database. https://github.com/absheikh/User-Authentication-With-Ktor-Freemarker-and-Postgres

The clickable "17" link in the latest blog post points to an empty list for me. @hhariri