https://kotlinlang.org logo
Channels
Powered by
#announcements
  • s

    stangl0r

    01/04/2017, 4:16 PM
    Ah I see. That might be, because Kotlin takes the best features of a lot of languages, also scripting-like languages.
  • s

    stangl0r

    01/04/2017, 4:16 PM
    No offense taken 🙂
  • r

    roberto.guerra

    01/04/2017, 4:18 PM
    ML also has first class functions that don’t require to hang off a class. Never occurred to me to see any similarity with JS. My first impression with Kotlin was: ooohhh, something like ML on the JVM. Finally.
  • j

    javad

    01/04/2017, 5:15 PM
    Has anyone had issues with Proguard striping kotlin.Unit object?
  • j

    javad

    01/04/2017, 5:16 PM
    The following rule doesn’t help, I guess because this is a standalone object rather than a class
  • j

    javad

    01/04/2017, 5:16 PM
    Copy code
    -keep class kotlin.Unit.** { *; }
  • j

    jw

    01/04/2017, 5:16 PM
    that would keep nested classes
  • j

    jw

    01/04/2017, 5:17 PM
    get rid of the 
    .**
  • j

    jw

    01/04/2017, 5:17 PM
    using 
    **
    in a proguard rule is almost never the correct thing
  • j

    javad

    01/04/2017, 5:17 PM
    oh cool! thanks Jake!
  • j

    javad

    01/04/2017, 5:18 PM
    so you suggest not using the ** wildcard in most cases?
  • j

    javad

    01/04/2017, 5:19 PM
    we also have another issue with Proguard striping top-level funcions in kotlin.Collections
  • j

    javad

    01/04/2017, 5:19 PM
    is there a way to keep them without keeping the whole package?
  • j

    jan.rabe

    01/04/2017, 5:21 PM
    i wish i wouldn't need to handle proguard rules myself, there needs to be a better way
  • j

    jan.rabe

    01/04/2017, 5:25 PM
    especially with tools out there like https://github.com/java-deobfuscator/deobfuscator
  • j

    javad

    01/04/2017, 5:36 PM
    @jan.rabe totally agree with you on the necessity of a better way for handling rules, but Proguard is as much about shrinking and method count reduction as much as it is about obfuscation. And even though there are de-obfuscation tools out there, you’re still slightly more protected, as far as I understand, since the naming is a great obstacle in code readability. no?
  • j

    javad

    01/04/2017, 5:53 PM
    Regarding my own question on keeping top-level functions in kotlin.collections.Collections.kt, there is a way: we can keep the 
    multifile
    java class 
    CollectionsKt
    that holds the functions/methods (https://kotlinlang.org/docs/reference/java-to-kotlin-interop.html) (and as @jw mentioned use the class name without .** wildcard):
    Copy code
    -keep class kotlin.collections.CollectionsKt { *; }
  • s

    skoude

    01/04/2017, 6:40 PM
    @rocketraman do you have any idea, can I get the Cursors hashMap values mapped to list of objects automatcally?
  • r

    rocketraman

    01/04/2017, 6:42 PM
    @skoude I found that Jackson does a very good job converting RethinkDB's maps into domain objects. For example, I had these three methods in my DAO base class:
  • r

    rocketraman

    01/04/2017, 6:42 PM
    Copy code
    protected fun typedValue(untypedValue: Map<String, Any?>?): T? =
    mapper.convertValue(untypedValue, klass.java)

  protected fun typedList(untypedIterable: Iterable<Map<String, Any?>>): List<T> =
    when(untypedIterable) {
      is Cursor<Map<String, Any?>> -> untypedIterable.use { it.map { typedValue(it)!! } }
      else -> untypedIterable.map { typedValue(it)!! }
    }

  protected fun typedSequence(untypedIterable: Iterable<Map<String, Any?>>): Sequence<T> =
    when(untypedIterable) {
      is Cursor<Map<String, Any?>> -> untypedIterable.use {
        it.iterator().asSequence().map { typedValue(it)!! }
      }
      else -> error("Expected cursor")
    }
  • s

    skoude

    01/04/2017, 6:45 PM
    @rocketraman, hmm.. I was somehow thinking that the Driver can do it automatically, so that's the reason I asked, but Jackson is a good solution for that. Thanks for the code sample.. Will continue my journey with kotlin and rethinkdb 😄
  • r

    rocketraman

    01/04/2017, 6:46 PM
    No the driver just deals in Maps. It doesn't concern itself with conversion to/from domain objects.
  • r

    rocketraman

    01/04/2017, 6:47 PM
    Going the other way (domain object -> map) is unfortunately more complex than just using Jackson, because the driver does not expect pure JSON, but a Map where some of the values are typed e.g. 
    OffsetDateTime
    for date/time values. I created a "mini-mapping" framework to handle these conversions.
  • e

    evanchooly

    01/04/2017, 6:47 PM
    didn’t rethinkdb die?
  • s

    skoude

    01/04/2017, 6:48 PM
    Yeap, the commercial side, but it was opensourced
  • r

    rocketraman

    01/04/2017, 6:49 PM
    The company died. The product is still open source. Looks like they are making some progress transitioning it to a non-commercial-entity backed open source model.
  • s

    skoude

    01/04/2017, 6:51 PM
    hmm.. it would be great if we would have kotlin driver that could make the conversions automatically..
  • r

    rocketraman

    01/04/2017, 6:54 PM
    That would be great. I chose to make my conversions from objects -> map explicit (with a custom DSL to make it super-simple) rather than using reflection and/or annotations. But it wouldn't be difficult to make it generic.
  • s

    skoude

    01/04/2017, 7:07 PM
    yeah the custom DSL is a good solution for this.. I think I will implement it like that also.
  • n

    nhaarman

    01/04/2017, 7:28 PM
    Is it possible to mark parameters in java libraries as non nullable? For example for RxJava, I'd like the compiler to fail if I try to pass 
    null
    to an 
    onNext
    call. I can annotate the method in Intellij with 
    null -> fail
    , but that doesn't seem to do anything
1...232223232324...3338Latest