slick/slick

Scala Language Integrated Connection Kit

naftoligug

@naftoligug:matrix.org

[m]

I did slick/slick#2250

I usually do slick stuff on sundays

in any case not much I can do until that is approved. So if anyone can review please do so ;)

Well, I suppose I could just work on a branch off of that branch.

Ok so ping me when you're on and I'll see if I can do some more stuff then

Antoine Doeraene

@sherpal

I'm there if you'd like to give it a go, otherwise I can do on Sunday :) (Yesterday it was Belgium-Croatia so I had to watch :p )

nafg

@naftoligug:matrix.org

[m]

can't now. Until what time?

Antoine Doeraene

@sherpal

now it's 9pm for me, until 11pm.

but otherwise on Sunday it's good for me :)

Naveen

@dexter2305

Model class & Table class are respectively. I am not able to understand compilation failure with insert Query. Can someone help ?

case class Task(id: Int = -1, text: String, project: Option[String], context: Option[String], completed: Boolean)

class Tasks(tag: Tag) extends Table[Task](tag, "tasks"){
  def id: Rep[Int] = column[Int]("id")
  def text: Rep[String] = column[String]("text")
  def project: Rep[Option[String]]= column[Option[String]]("project")
  def context: Rep[Option[String]] = column[Option[String]]("context")
  def isCompleted: Rep[Boolean] = column[Boolean]("isCompleted")
  override def * = (id, text, project, context, isCompleted) <> ( Task.tupled, Task.unapply)
}

InsertQuery looks like this now.

val insertQuery= tasks += newTask

A lot of examples are in similar way but I could not figure out the difference here for a couple of hours now.

Naveen

@dexter2305

Found: I missed import slick.jdbc.PostgresProfile.api._. Is there a way to simplify the inclusion ? class Tasks should have this and also the class that has the insertQuery. That factor is forcing to make Tasks & Repositoy to be in the same file.

nafg

@nafg

Nah you just need that import in every file you do query stuff

Scala 3 export might be able to solve that, not sure

but slick doesn't support scala 3 yet

Antoine Doeraene

@sherpal

However you can still abstract away which profile you use (it can be helpful if you want to use the H2Profile in your tests (especially since compatibility between PostgresProfile and the postgres mode for h2 broke)

nafg

@naftoligug:matrix.org

[m]

In general that's not a good idea because of inherent differences between databases, especially nowadays that it's so easy to spin up a postgres instance with docker or via testcontainers.org

Antoine Doeraene

@sherpal

Yes for advanced uses it's usually bad... But spinning a docker is still quite expensive

nafg

@naftoligug:matrix.org

[m]

It shouldn't be expensive

Expensive how?

Antoine Doeraene

@sherpal

I'm only speaking about computational power. You easily get 10-20s of overhead

1 reply

Naveen

@dexter2305

I am looking at "how to design the classes with domain object Task (as quoted in the snippet above) along with DAO/Repository. Now that we spoke of H2 (only for testing), question I am getting is - is it a good idea to test with H2 when the production is going to be something else. Kind of beginner with slick - not sure if this advisable. If so - any directions ? Thanks

Antoine Doeraene

@sherpal

As nafg said above, it's easy (in terms of code) these days to spin up a database inside a Docker container with your database of choice (in this case, postgres). The only thing that you have to do is to apply the evolutions/migrations to it (because obviously, you get a fresh one) and put some data in it. How you apply these evolutions will depend on your evolutions framework (on the JVM in scala the two most probable are Flyway or Play, let us know if you need help ;) ). For managing the docker container, as suggested also, you can use https://github.com/testcontainers/testcontainers-scala

Naveen

@dexter2305

Yes, I am spinning a docker container for postgres. Since it will be a fresh one, classSchemaGenerator is needed. SchemeGenerator requires Database and collection Table[Model]. There is also this sticky import statement import slick.jdbc.PostgresProfile.api._ which I want to keep it min files as possible. Given these I am looking for design these classes.

Antoine Doeraene

@sherpal

The question is why you want to keep that import statement in as min files as possible? If it's because your project is the barebones for building other projects, then you can always store somewhere in a global object the profile you want to use. Something like

object DatabaseGlobals {
  def profile = slick.jdbc.PostgresProfile
}

and then you can import that instead, via my.package.DatabaseGlobals.profile.api._.
If the reason is for mocks in tests, then I think you can put your table definitions inside a trait which requires an abstract jdbc profile, and feed that profile by extending the trait (I think that works)

Margus Sipria

@margussipria

I there easy way to make extension to Query that will allow add NOWAIT to select queries for Postgres?

nafg

@nafg

Yeah there's some method to modify the SQL, I don't remember offhand though

Margus Sipria

@margussipria

yah, i have tried to find out, but probably not well documented, haven't find anything

nafg

@naftoligug:matrix.org

[m]

@margussipria: I found something: https://stackoverflow.com/questions/65313912/customise-generated-slick-sql-for-debugging

Margus Sipria

@margussipria

this seems to work

    val action = query.result.headOption
    action.overrideStatements(action.statements.map(_ + " NOWAIT"))

i'm just worried if binding will happen correctly in every case. was wondering if there could be some way making this work with WrappingQuery (like forUpdate)

  implicit protected class AddNoWait[E, U, C[_]](val q: Query[E, U, C]) {
    def nowait: Query[E, U, C] = {
      ???
      new WrappingQuery[E, U, C](q.toNode, q.shaped)
    }
  }

nafg

@naftoligug:matrix.org

[m]

No idea sorry

Anil Singh

@asjadoun

Is there a way to connect to Sybase database ? Any help or example is appreciated.

Antoine Doeraene

@sherpal

There is a jdbc driver so in principle that means you can connect. The difficulty is that you need to teach slick the syntax via a profile, unless the syntax is the same as another existing db

Naveen

@dexter2305

The question is why you want to keep that import statement in as min files as possible? If it's because your project is the barebones for building other projects, then you can always store somewhere in a global object the profile you want to use. Something like
object DatabaseGlobals {
  def profile = slick.jdbc.PostgresProfile
}
and then you can import that instead, via my.package.DatabaseGlobals.profile.api._.
If the reason is for mocks in tests, then I think you can put your table definitions inside a trait which requires an abstract jdbc profile, and feed that profile by extending the trait (I think that works)

@sherpal - Intention is to use an in memory database for testing.

nafg

@nafg

I plan to do some maintenance tasks in say half an hour live on https://discord.gg/afDsPzmU, if anyone wants to join

nafg

@nafg

On now

Debasish Ghosh

@debasishg

Here is something that I am trying to do with Slick.

I have a column in a table which is NOT a primary key but I want to have a global ordering on it. Initial thinking was to have it an AUTO_INC column
On insert things are fine with the AUTO_INC doing its part in assigning an incremented value automatically
Now my use case demands that when a row gets updated I want to change the value of that column to the (max-value of the column + 1), something like update table set global_offset = select max(global_offset)+1 from table

Besides the fact that this might have performance implications, the underlying sequence of the AUTO_INC column does not get updated. Hence this same value can be inserted in another row as well through an insert statement.

Any help how I can do this with Slick ? Thanks.

nafg

@nafg

@debasishg what RDBMS?

I think you have a few different questions. I would suggest first figuring out how to do it in SQL in your RDBMS, then breaking down the different Slick questions that remain

In general, Slick does not have built in support for update expressions, AFAIK (only literal values)

Of course you can use the sql string interpolator, or construct DBIOs with raw JDBC interaction (I forgot the exact way to do that offhand)

Debasish Ghosh

@debasishg

@nafg Can I do the following in Slick ? I know AutoInc is a better option but I want to have a handle on the sequence as I would like to use it during update of this table.

CREATE TABLE state (
  offset integer NOT NULL DEFAULT nextval('state_ordering_seq'),
  ..
);

nafg

@nafg

@debasishg I'm not sure if O.Default takes an expression or raw SQL. But in general I don't recommend using Slick to generate your DDL

It's nice that you can write slick table definitions and get some CREATE TABLE statements but then how do you do migrations? IMO it's convenient for getting started but not practical for production code

Especially if you codegen your table definitions from the database

If you draw a dependency diagram between the database and your code the arrows shouldn't reveal a cycle

I use Flyway. Write your DDL in Flyway migrations, and then worry about how your code talks to the database

Jeremiah Malina

@jjmalina

Can someone tell me if this config looks correct for mysql using Slick 3.3.1 on Scala 2.12?

nc {
  mysql {
    profile = "slick.jdbc.MySQLProfile$"
    dataSourceClass = "slick.jdbc.DatabaseUrlDataSource"
    properties = {
      driver = "com.mysql.cj.jdbc.Driver"
      databaseName = "analytics_cache_schema"
      serverName = "localhost"
      portNumber = 3306
      user = "analytics_cache"
      password = "qwe90qwe"
      characterEncoding = "utf8"
      useUnicode = true
    }
    numThreads = 10
    keepAliveConnection = true
  }
}

I initialize with Database.forConfig("nc.mysql") but getting this error: java.lang.ClassNotFoundException: slick.jdbc.DatabaseUrlDataSource
It only happens in EMR 6.0.0 in Spark 2.4.4, and not in my tests run by sbt
If I inspect my jar with I can see slick/jdbc/DatabaseUrlDataSource.class is included

nafg

@naftoligug:matrix.org

[m]

Can you post the full output

1 reply

Naveen

@dexter2305

@jjmalina I used a similar configuration with mysql. I have build.sbt with "mysql" %% "mysql-connector-java" % "6.0.6". Can you check if you mysql driver library dependency defined in your build.sbt ?

1 reply

Where communities thrive

People

Repo info

Activity