I'm confused by the way the slick 3 documentation describes transactions. I have slick 2 code that looks like this:
def doSomething(???) = DB.withTransaction { implicit session =>
userDao.doSomething(???)
addressDao.doSomething(???)
contactDao.doSomething(???)
}
How can i span a transaction in slick 3?
Please have a look at the documentation here http://slick.typesafe.com/doc/3.0.0/dbio.html#transactions-and-pinned-sessions
The idea is that you wrap a sequence of IO operations into a transactionally like shown in this example:
val a = (for {
ns <- coffees.filter(_.name.startsWith("ESPRESSO")).map(_.name).result
_ <- DBIO.seq(ns.map(n => coffees.filter(_.name === n).delete): _*)
} yield ()).transactionally
val f: Future[Unit] = db.run(a)
This way Slick still process all the operations reactively, but it runs them all in one transaction sequentially.
So your example would look like this:
def doSomething(???) = (for {
_ <- userDao.doSomething(???)
_ <- addressDao.doSomething(???)
_ <- contactDao.doSomething(???)
} yield()).transactionally
val dbAction = (
for {
user <- userTable.doSomething
address <- addressTable.doSomething
contact <- contactTable.doSomething
} yield()
).transactionally
val resultFuture = db run dbAction
You just need to wrap your action into 'transactionally'. Slick would take care of running all wrapped DB actions as a transaction.
Apart from the standard advantages of having a more reactive/functional/async way of writing code, it allows couple of performance improvements. Like it can determine at runtime if multiple actions can use the same session or not. In Slick 2.0, whenever you use 'withTransaction' or 'withSession' that opens a new jdbc session while here it has potential to reuse the same.
Related
I'm trying to compose monads in Scala doing some requests to server.
here is the code snippet that I'm using. I try not to use flatmap as possible only using for comprehension as well. any ideas? I know using Monad Transformers, but I don't know how to compose multiple monads. can anyone help me out?
for {
session <- getSession(ticker) //IO[Future[Response]]
crumbF = session.flatMap(response => Future(parseCrumb(response.body)))
cookiesF = session.flatMap(response => Future(response.cookies))
crumb = Await.result(crumbF, 5 seconds) // Future[String]
cookies = Await.result(cookiesF, 5 seconds) //Future[Seq[Cookies]]
data <- getData(ticker, startDate, endDate, interval, crumb, cookies.head) // IO[Future[Response]]
stocksF = data.flatMap { response =>
import DefaultBodyReadables._
Future {
StockDf.mapDataToDf(response.body)
}
}
} yield stocksF
So a few things.
If you launch futures inside a for comprehension then they will run in sequence rather than in parallel - if this is your intention then fine. If not then instantiate them outside the for comprehension.
You cannot mix monadic contexts inside a for comprehension.
// Yes
for {
a <- Some(5)
b <- Some(10)
} yield 5 * 10
// No
for {
a <- Some(5)
b <- Future(10)
} yield 5 * 10
I am starting to work with slick and scala and it seems that I still do not have the basics iron out:
I am using slick 3.0.0 and scala 2.11.7 and I am connecting to an Oracle database.
I want to get the number of rows in my table so I went and did a search and found the following:
1) This one tells me that .run does not exist:
mytable.length.run
2)This one tells me that there is a type mismatch found: slick.lifted.Rep[Int] and expected String:
var q = for(u <- mytable) yield u.id
print(q.size)
3)This one compiles and runs but prints Rep(Pure $#309962262):
var q = for{row <- mytable} yield row
println(Query(q.length))
So I am not sure if it is because I do not understand how this works but I was imagining that the following should happen:
A) constructQuery
b) "run" query.
So the other queries that I am using are as follow:
val db = Database.forConfig("Oracle")
try{
val f: Future[Unit] = {
val query: StreamingDBIO[Seq[String], String] = participants.map(_.id).result // A)"Construct query"
val stremQuery: DatabasePublisher[String] = db.stream(query) //B) "Run query"
streamQuery.foreach(println)
}
Await.result(f, Duration.Inf)
}
finally db.close
What am I missing? is number 3 not giving me what I want because is not under a db.stream/db.run/db.something command? Or am I just lost =)
Thanks in advance
Tona
Querying a database in slick basically consists of these three steps:
Create a Query
Convert Query to an Action
Execute the Action on a database
So your example would look something like this (types are optional and added for clarity):
val query = mytable.length // length is an aggregation
val action = query.result
val result: Future[Int] = db.run(action)
// Access result in a non blocking way (recommended):
result.map(count: Int => ...)
// or for completeness use Await (not recommended):
val count: Int = Await.result(result, Duration.Inf)
Further reading:
Slick documentation, Queries
Futures and Promises in scala
Is it possible to update variable column list, which number is know only in runtime by slick 3.0?
Below is example what I want to do (won't compile)
var q: Query[UserTable, UserTable#TableElementType, Seq] = userTable
var columns = List[Any]()
var values = List[Any]()
if (updateCommands.name.isDefined) {
columns = q.name :: columns
values = updateCommands.name.get :: values
}
if (updateCommands.surname.isDefined) {
columns = q.surname :: columns
values = updateCommands.surname.get :: values
}
q = q.filter(_.id === updateCommands.id).map(columns).update(values)
Here is what I've done in Slick 3.1. I wasn't sure what worse, editing plain SQL statement or multiple queries. So I decided to go with latter assuming Postgres optimizer would see same WHERE clause in update queries of single transaction. My update method looks like this
def updateUser(user: User, obj: UserUpdate): Future[Unit] = {
val actions = mutable.ArrayBuffer[DBIOAction[Int, NoStream, Write with Transactional]]()
val query = users.withFilter(_.id === user.id)
obj.name.foreach(v => actions += query.map(_.name).update(v))
obj.email.foreach(v => actions += query.map(_.email).update(Option(v)))
obj.password.foreach(v => actions += query.map(_.pwdHash).update(Option(encryptPassword(v))))
slickDb.run(DBIO.seq(actions.map(_.transactionally): _*))
}
In Slick 3.0 they adopted slightly different approach, instead of having updateAll methods, as far as I userstand path of combinators was adopted.
So main idea is to define some actions on the data and then combine them ont he database to make a single run.
Example:
// let's assume that you have some table classes defined somewhere
// then let's define some actions, they might be really different
val action: SqlAction = YourTable.filter(_id === idToAssert)
val anotherAction = AnotherTable.filter(_.pets === "fun")
// and then we can combine them on a db.run
val combinedAction = for {
someResult <- action
anotherResult <- anotherAction
} yeild (someResult,anotherResult)
db.run(combinedAction) // that returns actual Future of the result type
In the same way you can deal with lists and sequences, for that please take a look here: http://slick.typesafe.com/doc/3.1.0-M1/dbio.html
DBIO has some functions that allows you to combine list of actions to one action.
I hope that idea is clear, if you have questions you are wellcome to the comments.
to update a variable number of columns you may use this way as I used for slick 3:
def update(id: Long, schedule: Schedule, fieldNames: Seq[String]): Future[_] = {
val columns = schedules.baseTableRow.create_*.map(_.name).toSeq.filter(fieldNames.map(_.toUpperCase).contains)
val toBeStored = schedule.withDefaults
val actions = mutable.ArrayBuffer[DBIOAction[Int, NoStream, Write with Transactional]]()
val query = schedules.withFilter(_.id === id)
//this is becasue of limitations in slick, multiple columns are not possible to be updated!
columns.find("NAME".equalsIgnoreCase).foreach(x => actions += query.map(_.name).update(toBeStored.name))
columns.find("NAMESPACE".equalsIgnoreCase).foreach(x => actions += query.map(_.namespace).update(toBeStored.namespace))
columns.find("URL".equalsIgnoreCase).foreach(x => actions +=
db.run(DBIO.seq(actions: _ *).transactionally.withPinnedSession)
}
I'm using Slick 3.0, and following is my codes:
def registerMember(newMember: TeamMember): Future[Long] = {
db.run(
teamProfileTable.filter(u => u.ID === newMember.ID).result.headOption
).flatMap {
case None => Future(-1)
case _ => db.run(
(teamProfileTable returning teamProfileTable.map(_.staffID)) += newMember.toTeamRecord
)
}
}
This may look ok. But when there are more layers of callback, the codes may become hard to read. I tried to simplify the codes using for-expression or andThen.. But due to the pattern matching part, I can only use flatMap to implement this..
Does anyone have ideas about how to refactor this?
I think a for comprehension should be okay here, you just need conditional handling of the Option in the result of the first Future. Something like this should work (note I did not compile check this):
def registerMember(newMember: TeamMember): Future[Long] = {
for{
r1Opt <- db.run(teamProfileTable.filter(u => u.ID === newMember.ID).result.headOption
r2 <- r1Opt.fold(Future.successful(-1L))(r1 => db.run((teamProfileTable returning teamProfileTable.map(_.staffID)) += newMember.toTeamRecord)
} yield r2
}
You can see on the right side of the fold that I have access to the result of the first Future if it was a Some (as r1).
I would even take this a step further and create separate methods for the steps of the for comprehension to clean things up, like so:
def registerMember(newMember: TeamMember): Future[Long] = {
def findMember =
db.run(teamProfileTable.filter(u => u.ID === newMember.ID).result.headOption
def addMember(r1Opt:Option[TeamMember]) = {
r1Opt.fold(Future.successful(-1L)){r1 =>
db.run((teamProfileTable returning teamProfileTable.map(_.staffID)) +=
newMember.toTeamRecord)
}
}
for{
r1Opt <- findMember
r2 <- addMember(r1Opt)
} yield r2
}
Another approach to simplify nested db.runs in Slick 3.0 when the query spans two tables could be to join the queries into a single query. Joining and Zipping. However, the OP seems to have the somewhat rarer case of nested queries on the same table so this approach may not be helpful in that particular case.
val query = slickLoginInfos join slickUserLoginInfos on
((l,ul) => l.id === ul.loginInfoId)
db.run((for { (l, ul) <- query } yield (ul)).result.headOption)
How do you 'un-lift' a value inside a query in Slick when using lifted embedding? I was hoping a 'get', 'toLong' or something like that may do the trick, but no such luck.
The following code does not compile:
val userById = for {
uid <- Parameters[Long]
u <- Users if u.id === uid
} yield u
val userFirstNameById = for {
uid <- Parameters[Long]
u <- userById(uid)
---------------^
// type mismatch; found : scala.slick.lifted.Column[Long] required: Long
} yield u.name
You can't, for 2 reasons:
1) with val this is happening at compile time, there is no Long
value uid. userById(uid) binds a Long uid to the compile time
generated prepared statement, and then .list, .first, etc. invoke
the query.
2) the other issue is as soon as you Parameterize a query,
composition is no longer possible -- it's a limitation dating back to
ScalaQuery.
Your best bet is to delay Parameterization until the final composed query:
val forFooBars = for{
f <- Foos
b <- Bars if f.id is b.fooID
} yield(f,b)
val allByStatus = for{ id ~ active <- Parameters[(Long,Boolean)]
(f,b) <- forFooBars if (f.id is id) && (b.active is active)
} yield(f,b)
def findAllByActive(id: Long, isActive: Boolean) = allByStatus(id, isActive).list
At any rate, in your example you could just as well do:
val byID = Users.createFinderBy(_.id)
The only way that I know to get this kind of thing to work is wrap the query val in a def and pass in a runtime variable, which means Slick has to re-generate the sql on every request, and no prepared statement is sent to underlying DBMS. In some cases you have to do this, like passing in a List(1,2,3) for inList.
def whenNothingElseWorks(id: Long) = {
val userFirstNameById = for {u <- userById(id.bind)} yield u.name
}