Office Hours — Today, July 27

← Thursday, July 25

Jul 27
8:50 AM
Mark M.
has entered the room
Mark M.
turned on guest access
8:55 AM
Paul R.
has entered the room
Mark M.
hello, Paul!
I'd ask "how can I help you today?", but I think that I know the answer... :-)
Paul R.
Good morning. I just sent you some email about my questions for today. I'll give you a chance to read the message.
Mark M.
regarding the SO question, "2" feels like the standard JVM way of naming anonymous inner classes
if you think of all those wonderful stack traces you get, where listeners and whatnot show up as numbered things
(BTW, for anyone reading this transcript, https://stackoverflow.com/q/57229715/115145 is the question)
this is the sort of Kotlin thing that might vary by environment (Kotlin/JVM vs. Kotlin/JS)
Paul R.
Correct which makes me thihk "this" is not what I think it is.
i.e. "this::class.simpleName"
Mark M.
"this" should be your anonymous instance of Human
ansh s.
has entered the room
Mark M.
but, on the JVM, that's just an object from an anonymous inner class of whatever class your function is in
(BTW, hello Ansh -- I will be with you shortly!)
Paul R.
Should be and if it is, then simpleName is not working per the spec or I don't understand something. I'm guessing the latter.
9:00 AM
ansh s.
hello, and no problem
Mark M.
ah, I see
when you say "spec", do you really mean the language specification, or just the KDocs?
Paul R.
KDocs, which I repeat in the question.
Mark M.
my guess is that it's a documentation bug
but, that's just a guess
Paul R.
The compiler and the docs are consistent fwiw.
Mark M.
I don't understand what you mean there, sorry
BTW, in Kotlin/JS 1.3.41, I get null for the simpleName
View paste 
interface Human { fun think(): String }

fun main() {
  fun start(man: Human) = println(man.think())

    start(object: Human {
        val name = this::class.simpleName
        override fun think() = "Thinking really hard! Name is: $name" // name == 2
    })
}
View paste 
output is:

Thinking really hard! Name is: null
you can try that for yourself in https://klassbook.commonsware.com/scratchPad.html
(which is a site that I'm still working on, so I apologize for any hiccups)
so, I'm back to "it might vary by platform (Kotlin/JVM vs. Kotlin/JS)"
and the docs might have been written with Kotlin/JS in mind or something
9:05 AM
Mark M.
let me take a question from Ansh, and I can follow up with you in a bit
Ansh: hi! how can I help you today?
Paul R.
For the same code you get a different result than I do.
ansh s.
So i was getting a bit confused about how to handle livedata with a room db having multiple tables and many to many relations
My UI model will have a Class A with {id, name and List<B>} where B is another class having {id, name}
Mark M.
LiveData itself only cares about your queries and their results -- it does not care about the number of tables or the nature of any relations
(technically, LiveData does not even care about the queries, as it is a data holder -- the Room InvalidationTracker cares about the queries and their results)
ansh s.
i have thus 2 database entities A and B and a relational entity AB. I handled the inserion part: adding id, name in A and all the Bs objects in B followed by their id-id relations in entity AB
But how can i return a livedata
My dao is now an abstract class
Mark M.
I assume that you are handling all of that work in a @Transaction method on your DAO
9:10 AM
ansh s.
yes
Mark M.
I don't recall @Transaction supporting much in the way of a return value, though
let alone LiveData
ansh s.
i can't return a value in a transaction?
Mark M.
you can, but Room isn't generating the code for creating the return value
you are on your own for that
ansh s.
yes, right, i remember being able to do that in another project, but this oe uses single livedata streams and now i have extended it to use multiple entities, i don't know how to get a livedata of that
this one*
Mark M.
you would need to create it yourself, I imagine
I'm not certain what data type you are planning on it being a LiveData of, though
insert operations return an ID value
so, your @Transaction method is getting a handful of ID values (for A, for B, for AB)
are you looking to return a LiveData of those IDs?
ansh s.
yes, all encapsulated in a completely new entity AB_ui ={A's Values , List<B> objects}
Should i use some better wordings for A/B class ? They are like A note and a tag. A note can have many tags
Mark M.
then isn't that a separate @Query method on your DAO, that your @Transaction method delegates to after doing the inserts?
9:15 AM
ansh s.
yes, for insertion , it does. i can make an abstact @Query method for getting A, B and AB resppectively too, But i don't know, weather they should return A/B/AB or LiveData<A>, LiveData<B>,...etc
Mark M.
LiveData does not have a zip() operator, though there is at least one open source one floating around
ansh s.
whats a zip operator?
Mark M.
zip() is a method you can use in RxJava to say "combine the outputs of these reactive types into a single object" which itself is delivered reactively
you would zip your A result, your B result, and your AB result into the single result representing that entire data structure
zip() would wait until all three of those results are in before emitting the combined output
ansh s.
and the result is itself a single stream?
Mark M.
yes
that's the sort of thing you would need if you wanted your @Transaction to return a LiveData<A_Plus_B_Plus_AB>
ansh s.
oh, then that's really what i want.
and thats not possible with room/livedata, right?
9:20 AM
Mark M.
this has come up before, and I know that there is at least one open source library that offers a zip() transformation for LiveData
9:20 AM
ansh s.
is it possible to use mutable livedata here?
Mark M.
though I think that the developer who brought all this up in chat wound up rolling his own zip() implementation
ansh s.
wow
Mark M.
MediatorLiveData is somewhat more likely, but you will need something like that
it's kinda a pain
Kotlin coroutines simplify this a ton
ansh s.
ohh
Mark M.
but poke around and see if you can find an existing zip() for LiveData in a library
let me take another question from Paul, and I'll be back with you in a bit
ansh s.
oh okay
Mark M.
Paul: back to you! what would you like to talk about next?
Paul R.
Back to my SO issue: It seem to work in JS for you but not for me and another person on JVM.
Mark M.
I am not terribly surprised
Paul R.
The other guy on the Kotlinlang Slack says it feels like a bug got introduced.
Mark M.
possibly -- as somebody pointed out on a related question that I had, "this is the sort of thing that I would expect would vary by platform"
Paul R.
I'm guessing it is a Java 7 -> Java 8 bug.
Mark M.
that's a question for JetBrains, I guess
Paul R.
Correct and I shall make it so. :-)
9:25 AM
Paul R.
Now if you would tell me about your attention to Kotlin vs Android and "bump"
Mark M.
"Are you interested in having us "bump:" Kotlin issues for you?" -- I'm not completely certain what you mean by "Kotlin issues"
if you mean literally things in their issue tracker, I think I have only ever filed one issue
Paul R.
The issue I posted to SO is not Android.
not Android related.
Mark M.
if you mean "can I use the Stack Overflow bump feature of the Warescription for pure Kotlin questions", yes, though I think I need to update my site to support that
I take a peek at the SO question tags and look for "android"
Paul R.
Yes, it is a tagging issue.
Mark M.
yeah, I've added a to-do item to support questions tagged "kotlin" as well as questions tagged "android"
Paul R.
I think you need your own tag. :-)
Mark M.
give me a week or two to get around to that
in the meantime, a workaround would be to temporarily add the android tag, use the bump, then remove the android tag
though the 24-hour waiting period rule would still apply
Paul R.
Yup.
Mark M.
but, probably next weekend if all goes well, I'll support the kotlin tag directly
thanks for pointing this out!
9:30 AM
Mark M.
let me take another question from Ansh, and I'll swing back to you after that
Paul R.
Are there any cases where, for a person having to learn both coroutines and Rx, you would say: learn both.
Mark M.
Ansh: back to you! do you have another question?
(Paul: if you have to maintain an existing Rx codebase, you'd need to learn enough Rx to do that -- otherwise, I think coroutines will be the long-term Android answer)
ansh s.
no , i was actually just thinking about what should i do with my app now, i actually have a very less time to push the new relations version
Maybe i will be removing livedata for now, i don't know rx java much
Mark M.
I can't really comment on that, as I know little about your app, its requirements, and your deadlines :-)
ansh s.
also, have you written something about custom views? there is a lot of mess and confusion in the official docs
Mark M.
there is a chapter on it in *The Busy Coder's Guide to Android Development*
but I only covered some simpler scenarios
https://wares.commonsware.com/app/internal/book...
it has never been documented very well
ansh s.
well, i will be looking into the book then, thanks!
9:35 AM
ansh s.
i am gonna go now, have a great day/evening/ night :)
Mark M.
you too!
Paul: back to you!
ansh s.
has left the room
Paul R.
I like your answer re: Rx. Music to my ears. And that does it for me. See you in a few weeks, btw.
Mark M.
very nice! see you then!
9:50 AM
Paul R.
has left the room
10:00 AM
Mark M.
turned off guest access

← Thursday, July 25

 

Office Hours

People in this transcript

  • ansh sachdeva
  • Mark Murphy
  • Paul Reilly