I updated from 1.5.4 to 2.0.0-alpha04 and the JVM version works as expected. I guess there's a bug in 1.5.4.

Hi there, Quick one in regards to dagger-hilt. I am getting a build exception when using @HiltViewModel annotation. This is the link to SO: https://stackoverflow.com/questions/73931588/injection-of-an-hiltviewmodel-class-is-prohibited-since-it-does-not-create-a-vi

It seems that Barber only supports rendering Strings, which is fair enough - does it have support for conditionals? If not, is it something that might be planned? Having a boolean field in my

DocumentData

doesn't seem to make it happy at all.

Just strings. Is that really two different documents?

Not two documents, rather, omitting or displaying something in the body of that single document

I'm generating font definition files, with a bunch of entries looking like this:

FontFace={ul}{n}{*-Thin},
  FontFace={ul}{it}{*-ThinItalic},
  FontFace={xl}{n}{*-ExtraLight},

Not every font has every weight (thin, extra light...) or shape (regular/italic) combination, so I'd like to do something like this:

{{#hasThin}}FontFace={ul}{n}{*-Thin},{{/hasThin}}
  {{#hasThinItalic}}FontFace={ul}{it}{*-ThinItalic},{{/hasThinItalic}}
  {{#hasExtraLight}}FontFace={xl}{n}{*-ExtraLight},{{/hasExtraLight}}

Barber uses regular Mustache under the hood so standard Mustache conditionals as you put are supported.

Maybe a bit of tangent for this channel but I'm using kotlinpoet to generate source code for a library I'm building. The library communicates with a service and the protocol is defined in a json file which I use to generate sources (request types, event types, enums, etc). However, I run into a circular dependency problem where the generated sources depends on the regular sources (e.g. implementing interfaces), but the regular sources would also benefit from depending on the generated sources (e.g. access to error codes enum, access to event types for deserialization). How can this typically be resolved? I don't have much past experience with generating sources like this. Atm, I'm relying on reflection in some few cases to deal with this problem but it's not particularly nice. Another approach would be to, instead of genering sources at compile time, generate sources on-demand and store them together with the regular source files. Would also mean that I'd have to check those into git.

How do I add a parameter to the constructor of a super interface?

addSuperclassConstructorParameter

doesn't seem to do anything

In Sqldelight 2.0.0, did anything change about the way migrations are run? For tests, we setup a sqlite tokenizing driver with

afterEvaluate {
        tasks.withType<VerifyMigrationTask>().configureEach {
            doFirst {
                TokenizingDriver(this.project).replaceJDBCDriver()
            }
        }
    }

This works fine pre-2.0.0, but now it throws the following exception:

> There was a failure while executing work items
   > A failure occurred while executing app.cash.sqldelight.gradle.VerifyMigrationTask$VerifyMigrationAction
      > Error compiling Error compiling CREATE VIRTUAL TABLE SearchText USING fts5(
            term,
            type UNINDEXED,
            sourceUid UNINDEXED,
            referredItemUid UNINDEXED,
            prefix = '2 3 4 5 6 7 8 9 10',
            tokenize = 'unicode61_synonym'
        )

Adding a logger to

DriverManager

provides us with the following Stacktrace:

org.sqlite.SQLiteException: [SQLITE_ERROR] SQL error or missing database (no such tokenizer: unicode61_synonym)
        at org.sqlite.core.DB.newSQLException(DB.java:1135)
        at org.sqlite.core.DB.newSQLException(DB.java:1146)
        at org.sqlite.core.DB.execute(DB.java:944)
        at org.sqlite.jdbc3.JDBC3PreparedStatement.lambda$execute$0(JDBC3PreparedStatement.java:56)
        at org.sqlite.jdbc3.JDBC3Statement.withConnectionTimeout(JDBC3Statement.java:429)
        at org.sqlite.jdbc3.JDBC3PreparedStatement.execute(JDBC3PreparedStatement.java:51)
        at app.cash.sqlite.migrations.CatalogDatabase$Companion.init(CatalogDatabase.kt:52)
        at app.cash.sqlite.migrations.CatalogDatabase$Companion.fromFile(CatalogDatabase.kt:36)
        at app.cash.sqlite.migrations.CatalogDatabase$Companion.withInitStatements(CatalogDatabase.kt:32)
        at app.cash.sqldelight.gradle.VerifyMigrationTask$VerifyMigrationAction.createCurrentDb(VerifyMigrationTask.kt:146)
        at app.cash.sqldelight.gradle.VerifyMigrationTask$VerifyMigrationAction.execute(VerifyMigrationTask.kt:119)
        at org.gradle.workers.internal.DefaultWorkerServer.execute(DefaultWorkerServer.java:63)

Using Paparazzi, is there some trick to get it to render the entirety of a

LazyColumn

without just setting the

ScreenHeight

to some arbitrarily large number? I have attempted to use

V_SCROLL

and

FULL_EXPAND

and neither seem to change anything. Additionally, the list it is rendering is immutable.

I tried the Zipline WorldClock Sample recently. When I tried to just make code changes to the server they were reflected on the app even when the app was open. I found this interesting, I thought that zipline would download the js code once when we call

zipline.take

and would cache that till the next call to

zipline.take

. Does the client poll the zipline server? (or some other way the server pushes data to the client like in grpc). Is there a way to ensure that even if the file is changed we don't make changes on the client side if the client is in the middle of a flow using zipline? Consider that the user is in the middle of a payment flow and we make changes to the js file it could it not lead to some unexpected behaviour?

I would like to select all possible values when empty list/null is passed to function or filtered values when selected filters are passed to the function. Recently I went through #1719 and it works great for single “id”:

WHERE (season = :season OR :season IS NULL)

Is it possible to achieve the same for collection of “ids” (with

IN

operator)? The following code doesn’t work, because it generates expected type as

Collection<Season?>

instead of

Collection<Season>?

and generally it’s not handled properly

WHERE (season IN :season OR :season IS NULL)

To overcome this issue I use the following statement:

WHERE CASE WHEN :seasonIsEmpty THEN TRUE ELSE season IN :season END

and it works great. However it forces me to introduce additional variable (

seasonIsEmpty

) which seems not optimal. I would expect that at least the following code should work:

WHERE CASE WHEN :season IS NULL THEN TRUE ELSE season IN :season END

but it also generates

Collection<Season?>

instead of

Collection<Season>?

Is there a way to simplify this statement?

What would be a good approach to run/verify paparazzi tests on my CI System? I always have normal non-paparazzi unit tests and paparazzi screenshot tests within the same module. Let's assume I have a gradle module,

sample

: •

./gradlew sample:verifyPaparazziDebug

would invoke the paparazzi tests as well as the other JUnit Tests. But I'd have to list all my modules in the project individually. •

./gradlew sample:testDebug

(or something like

./gradlew testDebugUnitTest

for the whole project) would run all the paparazzi tests as well as the other JUnit Tests, but it would not check if anything has changed (hence not fail). Is there any way to separate those two test groups? Does this even make sense? Or can I run everything (including the verification) in an easy way?

I think you're looking for the

check

task

./gradlew verifyPaparazziDebug

will run that task in all projects

you can use

./gradlew verifyPaparazziDebug testDebug

too, if you want only to run tests. If you want to run all checks, then

./gradlew check

I think

check

might also run different stuff (like linting), correct? I think I've tried

verifyPaparazziDebug

but it did nothing, since I didn't change anything in the source code 🙄 So I guess it was just cached.

I think this should work - thank you so much ❤️

Hi everyone, I'm using retrofit and I have the following question: Will retrofit able to parse one field of the response as generic? More details in the 🧵

Have a question on kotlinpoet or source code generation in general. I have a library project which generates sources. The code that generates sources depends on the regular sources. This can be solved by requiring the code generation to run after

compileKotlin

. However, this means that after first compilation of the project there aren't any compiled classes for the generated sources. I have a circular dependency where I can't run source code generation until classes are compiled but I also want the compile phase to include the generated sources. How to best deal with this? To me it would make sense if I could first compile specified classes needed to run the code generation, then run source code generation and lastly run the "regular"

compileKotlin

.

Is there a way to set HTTP trailers for a request in okhttp?

the OkHttp website is super broken, but https://square.github.io/okhttp/4.x/okhttp/okhttp3/-response/-builder/trailers/

Thanks Jake, that's on the response side though. Is there a way to set request trailers?

oh you're right. i assumed it was symmetric so only did a quick search. i only remember the PR land and again assumed it was symmetric.

I believe docs will be fixed on next release...

Hey all! I'm seeing this error when I try to upgrade to Kotlin 1.7.2:

e: java.lang.NoSuchMethodError: 'org.jetbrains.kotlin.ir.declarations.IrModuleFragment org.jetbrains.kotlin.ir.declarations.IrModuleFragment.transform(org.jetbrains.kotlin.ir.visitors.IrElementTransformer, java.lang.Object)'
        at com.squareup.anvil.compiler.ModuleMergerIr.generate(ModuleMergerIr.kt:41)
        at org.jetbrains.kotlin.backend.jvm.JvmIrCodegenFactory.convertToIr$lambda-1(JvmIrCodegenFactory.kt:121)
        at org.jetbrains.kotlin.psi2ir.Psi2IrTranslator.generateModuleFragment(Psi2IrTranslator.kt:88)
        at org.jetbrains.kotlin.backend.jvm.JvmIrCodegenFactory.convertToIr(JvmIrCodegenFactory.kt:146)
        at org.jetbrains.kotlin.backend.jvm.JvmIrCodegenFactory.convertToIr$default(JvmIrCodegenFactory.kt:64)
        at org.jetbrains.kotlin.backend.jvm.JvmIrCodegenFactory.generateModule(JvmIrCodegenFactory.kt:59)
        at org.jetbrains.kotlin.codegen.KotlinCodegenFacade.compileCorrectFiles(KotlinCodegenFacade.java:35)
        at org.jetbrains.kotlin.cli.jvm.compiler.KotlinToJVMBytecodeCompiler.generate(KotlinToJVMBytecodeCompiler.kt:331)
        at org.jetbrains.kotlin.cli.jvm.compiler.KotlinToJVMBytecodeCompiler.compileModules$cli(KotlinToJVMBytecodeCompiler.kt:123)
        at org.jetbrains.kotlin.cli.jvm.compiler.KotlinToJVMBytecodeCompiler.compileModules$cli$default(KotlinToJVMBytecodeCompiler.kt:58)
        at org.jetbrains.kotlin.cli.jvm.K2JVMCompiler.doExecute(K2JVMCompiler.kt:170)
        at org.jetbrains.kotlin.cli.jvm.K2JVMCompiler.doExecute(K2JVMCompiler.kt:52)
        at org.jetbrains.kotlin.cli.common.CLICompiler.execImpl(CLICompiler.kt:92)
        at org.jetbrains.kotlin.cli.common.CLICompiler.execImpl(CLICompiler.kt:44)
        at org.jetbrains.kotlin.cli.common.CLITool.exec(CLITool.kt:98)
        at org.jetbrains.kotlin.daemon.CompileServiceImpl.compile(CompileServiceImpl.kt:1618)
        at jdk.internal.reflect.GeneratedMethodAccessor108.invoke(Unknown Source)
        at java.base/jdk.internal.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
        at java.base/java.lang.reflect.Method.invoke(Method.java:566)
        at java.rmi/sun.rmi.server.UnicastServerRef.dispatch(UnicastServerRef.java:359)
        at java.rmi/sun.rmi.transport.Transport$1.run(Transport.java:200)
        at java.rmi/sun.rmi.transport.Transport$1.run(Transport.java:197)
        at java.base/java.security.AccessController.doPrivileged(Native Method)
        at java.rmi/sun.rmi.transport.Transport.serviceCall(Transport.java:196)
        at java.rmi/sun.rmi.transport.tcp.TCPTransport.handleMessages(TCPTransport.java:562)
        at java.rmi/sun.rmi.transport.tcp.TCPTransport$ConnectionHandler.run0(TCPTransport.java:796)
        at java.rmi/sun.rmi.transport.tcp.TCPTransport$ConnectionHandler.lambda$run$0(TCPTransport.java:677)
        at java.base/java.security.AccessController.doPrivileged(Native Method)
        at java.rmi/sun.rmi.transport.tcp.TCPTransport$ConnectionHandler.run(TCPTransport.java:676)
        at java.base/java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1128)
        at java.base/java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:628)
        at java.base/java.lang.Thread.run(Thread.java:829)

this only happens sporadically on some gradle targets. Running Kotlin 1.7.20, Anvil 2.4.2, dagger - 2.38.1. Double checked that if I remove the anvil plugin from the target, it'll compile just fine.

The bump to Kotlin 1.7.20 hasn’t been published yet https://github.com/square/anvil/pull/643 It seems like there’s a binary incompatibility in the Kotlin compiler. I suggest using the latest snapshot.

Multiplatform (coroutines) okio/okhttp question... A strong case is made for the blocking I/O model in the comment thread of this okio issue. The main argument being that dispatching once to

<http://Dispatchers.IO|Dispatchers.IO>

at the appropriate level to do your blocking read/write will be the most efficient. When we benchmarked using

suspend

for low level JSON tokenization in the AWS SDK for Kotlin we found this to hold true.

suspend

at too low a level (inner loops, small byte reads, etc) causes significant performance degradation. My question then is what about JS (node) support? The recommendation to block a thread is fine for JVM/Native, what would the equivalent be for JS?

In the one place I use it I buffer the whole body before wrapping