Yes. We are aiming for this week.

I am writing an integration for my android app view model calls into a repository which uses retrofit to make a network call (suspend functions) and then I use turbine to test the output to a StateFlow. My issue is that retrofit is making the call asynchronously and so the test completes before the call completes. How can I either (a) make the call synchronously or (b) somehow pass the TestCoroutineDIspatcher to OkHttp?

Hi 👋. Are there any known issues on

sqldelight

regarding Gradle caching? Seems that making kotlin compile depend on the generation task is causing wrong cached builds on my side.

tasks.withType<KotlinCompile> {
  dependsOn("generateMainSubscriptionsDbInterface")
}

Having that in place then running

./gradlew clean build

a few times, files are generated fine the first time, then they are removed from second time on, and they don't get regenerated ever, neither by running the generation task manually. It only gets back to life after removing the

.gradle/

dir. If I run the task manually after that, it generates the stuff again. I've also observed that the idea plugin allows to generate automatically after editing

.sq

files, but that is not enough for someone cloning the repo the first time or for someone not touching those files ever. Thanks!

And the IDE plugin is never used as the source of truth for compilation

Does SqlDelight support having multiple Database instances opened from different processes? I got asked this question and answered that it's

SqliteOpenHelper

stuff so it shouldn't be specific to sqldelight, but then I got pointed to some special handling made by Room in this regard:

RoomDatabase.enableMultiInstanceInvalidation()

. This has a side-effect of working correctly without "database is locked" exceptions. Is there anything like this in SqlDelight?

that would come down the driver, and the sqldelight packaged android driver just delegates everything to the support framework apis

So I thought right 🙂 It's just that Room seem to introduce something on top of framework apis for this multi-process case (see here), so I thought maybe sql delight has something too.

Hi 👋 Is

ON CONFLICT

working for postgresql dialect in sqldelight?

In SQLDelight can

INSERT

and

SELECT

statements be grouped? For instance, to return the ID of a newly inserted row.

@Derek Ellis yep, I think you can use this:

insert {
  INSERT INTO ...
  SELECT last_insert_rowid();
}

or

SELECT changes();

if you want the number of rows affected by an update or delete

It looks like the generated function returns

Unit

even with the select statement, and I just found an issue on github for this that I had missed 😳(https://github.com/cashapp/sqldelight/issues/2466)

oh whoops, yea that’d be useful to have

Hi there, do you plan to release new version of SQLDelight ? I would like to use it with KMM app with Kotlin 1.5.* . Or is there a snapshot?

The released version will work with Kotlin 1.5 just fine

But yes there are also snapshots of the development version

Hi! 👋 Are the mysqldelight

CREATE TABLE

postgresql statements expected to create tables automatically? I'm running a backend app and getting errors like:

subscriptionsdb | 2021-07-14 12:30:42.092 UTC [56] ERROR:  relation "userentity" does not exist at character 15
subscriptionsdb | 2021-07-14 12:30:42.092 UTC [56] STATEMENT:  SELECT *
subscriptionsdb |       FROM userEntity
subscriptionsdb |       WHERE slack_user_id = $1

as if it wouldn't be able to find the

userEntity

table.

CREATE TABLE IF NOT EXISTS userEntity (
    user_id SERIAL PRIMARY KEY,
    slack_user_id VARCHAR NOT NULL,
    slack_channel_id VARCHAR
);

insert:
INSERT INTO userEntity(slack_user_id, slack_channel_id)
VALUES (?, ?);

selectAll:
SELECT *
FROM userEntity;

select:
SELECT *
FROM userEntity
WHERE slack_user_id = ? AND slack_channel_id = ?;

selectByUserId:
SELECT *
FROM userEntity
WHERE user_id = ?;

selectBySlackUserId:
SELECT *
FROM userEntity
WHERE slack_user_id = ?;

selectBySlackChannelId:
SELECT *
FROM userEntity
WHERE slack_channel_id = ?;

delete:
DELETE FROM userEntity
WHERE user_id = ?
RETURNING *;

How do you usually aggregate in list in sqldelight? For example I have table of

Files

and table of

Folders

and I want to select all files and for each a list of folders to which it belongs. Something like this "pseudo" query:

SELECT file.id, file.name, COLLECT_IN_LIST(folder.name) as List<String>
FROM file
JOIN folder on folder.fileId == file.id
GROUP BY folder.name;

And have

data class Result(id: String, name: String, folders: List<String>

. But no COLLECT_IN_LIST aggregate function exists...

two queries. the first gives you the list of files, the second you can run for each file to get its list of folders.

That was one of the options here, but we also thought, who knows if there's some convenient syntax/feature we don't know about. thanks!

Not sure which version of sqlite you're using but there's JSON_GROUP_ARRAY.

There's GROUP_CONCAT if you're clever.

Thank you, will read up on them!

Hi all - I’ve been trying to understand how leak canary installs a seemingly separate app, just by including it as a dependency. The FAQ mentions the

leakcanary-object-watcher-android

module. I took a look through that and it seems like it is responsible for creating all the watchers (activity, fragment, VM, etc.), but what I don’t understand is how a separate LeakCanary application ends up on the device as well the project has quite a few modules so I would appreciate some pointers 🙂

Wanted to get some more eyes on this. Is there a simple way to grab a the discriminator name in a polymorphic fallback? My team is simply trying to log when we use the fallback, but it would be helpful to see what type exactly we failed to parse. https://stackoverflow.com/questions/68629197/moshi-polymorphicjsonadapterfactory-is-it-possible-to-get-the-type-in-withdefau/68643097#68643097

Hi guys - I’m using

moshi

in my app and I need to deserialize an escaped json string. I want to do this in a

@FromJson

adapter, but I’m not so sure how to do that. e.g. the raw json i have to deserialize looks something like this where the

metadata

is an escaped json string. so i want to make an adapter for the

metadata

object specifically

{
  "time_zone": "America/New_York",
  "metadata": "{\"foo\":\"bar\"}"
}

Currently not using neither Dagger nor Anvil, but plan to experiment with this combination. I wonder this: does using Anvil mean no kapt or is it still needed? Obviously I should dive in and find out, but I thought someone could answer quickly (didn't find related info in the README).

no kapt

wooow. priority raised 🙂

Kapt is still needed in the module where you use

@Component

or

@MergeComponent

, but in no other module. That would ideally be the app module.