Sqldelight: When the query returns only 1 column, should the compiler generate a new class with only one attribute or return the attribute instead without a (useless) wrapper class? In the gif from the website, after updating

playerForNumber

, the attribute

name

is not found.

When I use SQLDelight (1.5.3) I do

database.Schema.create(driver)

. I'm using a file based database. Second time I start my application the database already exists. I don't want it to recreate the tables. What is a good approach for that?

For SQLDelight with the MySQL dialect - is there any way to insert or update? i.e. it doesn’t seem like replace is supported. Am I missing something maybe?

Not creative but create tmp table, insert into tmp from selection * from current, drop current, rename tmp to current

@alec What would you recommend I do to understand whats going on in sql-psi so I can help contribute to the project. I understand kind of what’s going on in parts of the bnf files, but when I search online for what the heck bnf files are I don’t find anything relevant.

yeah, unescaping only. I had wrongly assumed that it would be a big task but it seems those four strings are the only cases I need to handle for my use case

We tend to use Compose as much as possible, and as a whole we believe it's the future for Android views. If one cannot write composable views, Contour is amazing.

Hey I am working in retrofit in android kotlin. I want to track progress when we uploading file in server. I tried this solution it's working fine.

implementation "com.squareup.retrofit2:retrofit:2.9.0"

1. How to pause uploading the file through server as well. 2. How to cancel uploading the file.

Hm.

val request = chain.request().newBuilder().removeHeader("If-None-Match").build()

Didn't work as an application interceptor. Going to give network interceptor a whirl

Found

@SingleIn

in Anvil's sample. I wonder will dagger differentiate

@SingleIn(Scope1::class)

and

@SingleIn(Scope2::class)

during code generation (and after) as different scopes? JSR docs seem to mention that

@Scope

-ed annotations shouldn't have attributes, so I thought does this mean dagger will look only at annotation name?

I'm seeing this error from applying anvil plugin:

e: java.lang.NoSuchMethodError: 'org.jetbrains.kotlin.ir.util.IdSignature$PublicSignature org.jetbrains.kotlin.ir.util.IdSignature.asPublic()'

found the similar github issue, it says to update anvil version, but I'm already using: kotlin-1.6.10 anvil-2.3.11 dagger-2.39.1 // this one is a bit old could it be because of dagger version?

I am using Anvil and I have a class with no arg constructor and which doesnt extend any interface and I'm tempted to add it with

@ContributesBinding(...)

, but this requires that target class extends some interface. Is it wrong to wish this? It just looks simpler than writing a boring provide with

MyClass()

in the Module.

Is there a way to force the order of Anvil `CodeGenerator`s so i can use a generated provision module in a separate generated file?

Why My

SQL Delight's

Todo notes app are not persisting data when I install My

Desktop Compose  App

as

DMG

file on my

M1 Macbook

? But everything works fine on my dev machine.

What could be the cause of getting (latest anvil version for 1.6.10):

com.squareup.anvil.compiler.api.AnvilCompilationException: Back-end (JVM) Internal error: Couldn't resolve class for anvil.module.AppComponentAnvilModule.
File being compiled: (11,1) in /anvil/module/AppComponentAnvilModule.kt

Any idea why IDE doesn't see the component contributed with anvil? You can see it's gray and unused (also there are red unresolved ref errors in usage sites), but it compiles fine. I can call

myComponent.createCoordinator()

and the instance is created. If I switch to non-anvil dagger it starts to work. Do I need to include some additional source set? Currently default from kotlin plugin are used

I’m considering using SQLDelight in a (multiplatform) project to store some read-only JSON documents. I was planning on implementing `ColumnAdapter`s that use

kotlinx.serialization

to serialize/deserialize an actual class from a String column I’m trying to determine how to handle changes to the JSON document’s schema (which is just a change in the

@Serializable data class

). In most of those cases, I’d be fine with dropping the old data and repopulating it next time I fetch from the server. I could write a

.sqm

file that drops all rows in the tables containing impacted documents, but I’m foreseeing situations where someone updates the data class without writing a migration. Since the sqlite schema would still just have the row as a String, I’m assuming that SQLDelight’s migration verifications would still pass. Is there a way to ignore/drop data that doesn’t deserialize properly in my

ColumnAdapter

? Should I just be making all of my types nullable?

Hi, I need to upload a jpeg to my server which accepts multipart(?). The image was taken with the camera on mobile (examples around the web reference browsers) and is held in memory. How to upload with retrofit? The docs talk about

RequestBody

but not sure how to put it all together.

hi all, when using paging on sqldelight, how can we handle query filtered?

countPlayers:
SELECT count(*) FROM hockeyPlayer WHERE full_name = ?;

players:
SELECT *
FROM hockeyPlayer WHERE full_name = ?
LIMIT :limit OFFSET :offset;

val pagingSource: PagingSource = QueryPagingSource(
  countQuery = playerQueries.countPlayers(full_name),
  transacter = playerQueries,
  dispatcher = <http://Dispatchers.IO|Dispatchers.IO>,
  queryProvider = playerQueries.players(full_name), // <---- issue
)

it seems like they're simply ignoring HTTP cache headers entirely and assuming everything can be cached forever. not sure that's a smart idea.

And.... we should fix

Does anyone have any good resources for writing tests for custom Anvil CodeGenerators?

Hi, in the moshi docs it state to use this version of ksp

id("com.google.devtools.ksp").version("1.6.0-1.0.1")

, this led to issues with codegen not working. I found the error message when building on the command line (gradlew build) but NOT when compiling and running my app with the AS. The app crashed at runtime saying the Adapter class could not be found. I found the latest version to use was

1.6.10-1.0.4

and it all worked.

I think https://square.github.io/okhttp/4.x/okhttp/okhttp3/-authenticator/ describes it well enough. for pre-emptive proxy auth, null continues with original request. for reactive auth, the original request was already made and came back with a unauthorized response; null makes that the final response instead of making a follow-up request

I'm trying to use

DATE_ADD

(https://dev.mysql.com/doc/refman/8.0/en/date-and-time-functions.html#function_date-add) in an sql statement with sqldelight, but sqldelight complains

'(', ')', '.', <binary like operator real>, BETWEEN or IN expected, got '6'

The used SQL code:

date_add(current_date() , interval 6 month)

Am I doing something wrong or is there a workaround for such a case?

Yep

Hi all on the version 2.x, how to use named query instead of anonymous

?

DELETE FROM customer WHERE customer_id = :customerId;

It doesn’t look working for this version but was working on version 1.x

This message was deleted.

Hey, I created a new .sq file, which has CREATE TABLE LastModificationDate(lastModificationDate TEXT NOT NULL); This causes a crash when testing against our production app as it says LastModificationDate. no such table: LastModificationDate: , while compiling: SELECT * FROM LastModificationDate I'm not sure how I am meant to migrate this new table, in the .sqm file I don't think I'm altering anything as its actually just a new table, so should I be creating the table in the .sqm? If so, how does that work with the functions that exist in LastModificationDate such as insert/get/delete