We have a use case where we want to watch for changes to four GraphQL queries, and when any of them change, we want to create and return parameters to use to make another query. In most cases adding GraphQL watch() to these works great. But there are a few cases where the cache doesn't have anything in it yet, and so I guess since there's nothing to watch() the code for making the parameters never fires. I'm stubborn, and want this to work. Any ideas on how to ensure at least one query gets made when initiating a watch()?

Hey got a question about testing. Suppose I have a simple UseCase, that receives an apolloClient in its constructor.It has one method that calls one query, and does some business logic with it and returns something. Is there no way to simply mock the ApolloClient to just return a specific object? I’ve read this article for example but it feels quite involved for what I want to do. Optimally I’d like to be able to do inside a test class something like:

private val mockedApolloClient = mockk<ApolloClient>().apply {
    val apolloClient = this
    coEvery { apolloClient.query(MyCustomQuery()/* or even omit this if possible?*/) } coAnswers {
        MyObject()
    }
}
private val useCase = MyUseCase(mockedApolloClient)

@Test
fun someTest() {
    assert(useCase(input) == SomeResult)
}

I tried looking into the documentation and I didn’t find anything out unfortunately. Note that we’re still on 2.x.x but if a better solution exists for 3.0.0 I’d be happy to hear that as we’ll migrate to it soon anyway.

It's not resolving

watch

below for some reason....is there some other dependency that needs to be pulled in for that? (basing code on what's in https://www.apollographql.com/docs/kotlin/caching/query-watchers/)

apolloClient.query(GetSessionsQuery())
            .watch()
            .collect { response ->
            }

Are there any particular recommended patterns/approaches when using

watch

with endpoint that supports paging?

APQ should only be attempted for queries, right? It seems as though the interceptor gets applied to mutations as well.

Hello! I am very excited about the new Apollo Kotlin, and I was looking at what it would take to upgrade the project from work from v2 to v3, but unfortunately it seems like a migration would be extremely difficult, from what I could tell so far. Even if using

codegenModels

set to

compat

, or with `useVersion2Compat()`(which I think that it might do the same thing), there would be thousands of changes to go through in the project, the vast majority of them would be just removing function invocations from GQL fields, quite a few of them would be changing how polymorphic fragments are accessed, and a lot of mutation and query builders need to be changed too to constructor invocations. It would be a huuuuge time sink to do all of this changes in one go, so I was wondering, is it possible to have v2 and v3 running side by side in the same app, but using the same SQL cache? So that we can keep the old code using Apollo-Android, and slowly migrate to Apollo-Kotlin. (This would actually solve another issue for us, we have a lot of custom code that allows us to pass references to fragments from one activity to another and then observe for updates, and we would have to rewrite most of it, but again, it's hard to do it in one go. Thankfully,

ApolloStore.readFragment()

from v3 does a lot of this already 😄.) I imagine that even if it would be possible to have two instances writing to and reading from the same database, observing the cache for updates using

.watch()

won't trigger any updates if the write was done by the other instance, isn't it? Is there any hope to ease the migration for such a complex project? I want the new and fancy test builders, haha. (Sorry for the wall of text.)

I'm using

SqlNormalizedCacheFactory

and for JVM have following (injected using Koin)

single { SqlNormalizedCacheFactory("jdbc:sqlite:swapi.db") }

Running a

main

function in JVM code and this works fine first time but on subsequent runs I get

Caused by: org.sqlite.SQLiteException: [SQLITE_ERROR] SQL error or missing database (table records already exists)
	at org.sqlite.core.DB.newSQLException(DB.java:1012)

Is there some other init/config needed for case of JVM?

Are there any examples of unit test that mocks out Apollo client?

Somewhat related (and this is very likely issue on my side but just in case anything jumps out)....I have a unit test in

commonTest

in KMM module using new

runTest

provided in Kotlinx Coroutines 1.6 to test some code that uses apollo...it runs fine for android, jvm but get issue when running for iOS .🧵

Should call like following be "main-safe"....or is expectation that client is on background thread when invoking it?

apolloClient.query(query).execute()

Screen Shot 2022-01-02 at 8.53.30 AM.png

Is it possible to handle network connection issues when using a

watch

? You said they watchers never throw here. Is it possible to emit network connection issues?

Could any guidance be added for interceptors, but not for Http ones as described here, but adding a section about ApolloInterceptor too. More in Thread 🧵

A random curiosity question about the gradle plugin 🧵

Is it not possible to use the test builders in non-test code? It is now generated in the test resources only. For our app we basically have a module called testdata where we have a ton of convenience builders (basically what test builders do, but we had to write them by hand which would be great if we could skip that from now on) which we use to make fake data when we test our app in some sort of “dev” mode. I was hoping we could use this functionality instead. Is there a way to generate them not only for the test sources?

Hello all! I have been updating a project to use Apollo and really liking it so far. I have a few questions about the test builders which are available in v3: 1. Is there an eta on when test builders will be out of experimental? 2. Do I always need to start from Query.Data when using test builders? In the example usage provided, it looks like you have to start with

HeroForEpisodeQuery.Data

, but it would be nice to be able to reference

hero

directly, especially in cases where we have more nested models. 3. Related to the previous question, how do I mock fields which are lists if I want to set some values of the list items but not all? Would I need to use real models?

Dealing with File uploading migration 2.0->3.0 tl;dr -> Before we were passing in a path string to FileUpload(mimeType, path) but now I’m not sure what’s the functionally equivalent thing to do. We used to do:

// graphqls
mutation UploadFile($file: Upload!){
  uploadFile(file: $file) { //stuff }
}
uploadFile(file: Upload!): File!
type File { // stuff }

// kt
private suspend fun uploadFile(path: String, mimeType: String): Response<UploadFileMutation.Data> {
    val uploadFileMutation = UploadFileMutation(
        file = FileUpload(mimeType, path)
    )
    return apolloClient.mutation(uploadFileMutation).execute()
}

and it’d work. Mind you, we were calling this method once with a file that we did

.path

on it, and once on a

Uri

that we did

.path

on it too. (so not all of our cases can be fixed with the Now with the new documentation mentioning “okioSource” as someone who has never worked with okio it’s not obvious what I should do to not break this. Maybe worth a mention in the migration logs? At least a link to okio documentation on this?

Hello 👋 Is it possible to attach an arbitrary object as a

tag

to an Apollo call, so that the tag can be read in an

okhttp3.Interceptor

class? The goal is to pass an object when triggering an Apollo call, and read it later from

okhttp3.Request#tag()

in the interceptor. There’s https://github.com/apollographql/apollo-kotlin/issues/2527 so I assume there isn’t (though maybe something appeared in v3?) and that the only way to do that now is to set a request header, read it in the interceptor, and strip before sending to the server?

Hello, do subscriptions work in multiplatform environment? Especially on iOS? Is there any public GQL server that supports subscription to test that out?

A question about more about coroutines and cancellations in combination with Apollo. More in thread 🧵

Using Apollo subscriptions for first time so could easily be issue on my side but am getting

java.net.ProtocolException: Expected HTTP 101 response but was '404 Not Found

for some reason even though subscription seems to work in playground....more in thread....

Is the normalized cache supported when using Kotlin/JS?

Hi folks, I'm the one who opened this ticket: https://github.com/apollographql/apollo-kotlin/issues/3756 cc : hi @bod

Hey, i’m trying to modularise the apollo queries within a mono-repo but getting duplicated class issue, currently this the structure i have:

monorepo
  app-a module
    apollo-a module
  app-b module
    apollo-b module

Currently what i get is:

Cause: duplicate Type XXX generated in modules: apollo-b,apollo-a

What i have tried is this: within app-a:

dependencies {
    implementation(project(":apollo-a"))
    apolloMetadata(project(":apollo-a"))
}

With apollo-a module:

apollo {
    generateApolloMetadata.set(true)
    generateKotlinModels.set(true)
}

Any suggestions on this? Apollo version: 2.5.9

Hi, I received the following duplication error for Apollo generated types, wondering where to add the 'alwaysGenerateTypesMatching' config?

Execution failed for task ':checkServiceApolloDuplicates'.

duplicate Type 'ResolverKey(kind=SchemaType, id=Query)' generated in modules: Somthing,SomethingElse

Use 'alwaysGenerateTypesMatching' in a parent module to generate the type only once

Did you guys run into any considerations for dropped connection and recovering from errors when doing sus over websockets? Specially but not limited to handover from mobile to wifi ?

I'm coming closer to the point of starting to write code on

apollo-kotlin

, though I have some server code that not sure where to put. At this point I'd prefer the server to run standalone, so that excludes putting all its code into

src/test

. Maybe a new module in

apollo-kotlin

, like

apollo-server

? But at the same time I don't have the goal of having a fully fledged server that actually handles true queries/mutations in a nice API (though that would be handy for my own goals if it were easy). Thoughts? Btw I'd like to have a quick voice chat during your office hours, where are you located? If US West Coast I can do something in my early hours, if Europe I can do it in late hours.

Is it possible to add a header on the introspection? We need to add an auth token for ours to work.

How might custom exceptions be thrown from an okhttp interceptor? I would like the exceptions to surface past all of the interceptors so that they can be handled elsewhere. More in thread ->

More of a project structure and IDE question I guess, but is there a way for me to share my .graphqlconfig without gitignoring it yet still have fields inside there that are hidden somehow? I just don’t quite know how to do that and any pointers would be appreciated!