FWIW if you have the choice I’d never use a compound primary key. You can have a compound unique index and it just makes life easier to have a single primary key. Also easier for the engine to optimize for.

I asked about passing value classes to retrofit functions on SO a while back (https://stackoverflow.com/questions/68232128/is-it-safe-to-pass-a-kotlin-inline-class-to-a-retrofit-function), and I want to verify that a few of my assumptions are correct here before updating SO: 1. Function mangling won't cause any issues because the call site will be updated to use the mangled function name 2. If the wrapped type is not boxed, a converter for the wrapped type will be used if one exists, otherwise

toString()

will be invoked on the wrapped type 3. If the wrapped type is boxed, a converter for the value type will be used if one exists, otherwise

toString()

will be invoked on the value type

wow what a spectacular answer by that person

i honestly have no idea what will happen though

your guesses seem right but i would simply have to try and see what happens to know for sure

it all sounds reasonable but I don't know how stable the ABI will be. what if a future version of Kotlin adds bridge methods for the boxed type (if unboxed) or unboxed type (if boxed)?

I just wrote some scenarios to test it, and that seems to be the behavior. I guess what is really making me nervous is what @ephemient said

They can only add bridge methods for new scenarios as changing existing ones would break binary compatibility

I wouldn't worry about it

I wouldn't worry about it much either, but the last time name mangling changed (

-Xuse-14-inline-classes-mangling-scheme

) Kotlin was compatible with either ABI as it used the metadata. was still an ABI break though.

sqldelight just does this for all drivers already

oh

just noticed the dollar signs

bizarre that a driver would require that

you can probably safely replace " ? " with " $ "

Generally the question marks are to prevent SQL injections. One would need to be extremely careful when doing

$

string formatting of SQL queries.

I’m currently using turbine. Working with an infinite flow I have to assert that no other event is emitted (in a reasonable amount of time). I’m not able to find a method that fits the purpose. I was thinking that the logic could be the same as

awaitItem

, but returning true in case of

TimeoutCancellationException

.

timeouts are gone from

trunk

as

runTest

now provides test-wide idle timeouts

Forgot to mention, with Kotlin 1.5.x 🙂

in reality there is no good way to test this

in our own tests we wait 1 second on wall clock time and assume that means we would wait forever

hi all, does ios support

app.cash.sqldelight:sqlite-3-38-dialect:2.0.0-alpha02

?

Thinking about rewriting a java library that only support File to internally works with Okio and support other things like urls. Considering that I might publish it, should I fully embrace Okio in the public API and only expose functions that accept Source/Sink or keep accepting File/InputStream/... or both ?

any Square people here know if there’s a copy of jrod’s Droidcon talk video somewhere other than the Vimeo one here? https://www.droidcon.com/2021/11/17/keeping-your-pixels-perfect-paparazzi-1-0-2/ asking because that copy starts out fine and loses audio sync as the video goes on 😕 can’t find a copy on youtube or anywhere else

If you use youtube-dl and play it with VLC locally you can fix such things. But yeah it'd be nice to have a working copy somewhere

funny you should mention that, I used youtube-dl to see if it was still out of sync (which it was) but I didn’t think to use VLC to fiddle around with the timing. I’ll give that a try, cheers

Hi, Does cashapp/paparazzi work with robolectric? When I use @RunWith(AndroidJUnit4::class) with Robolectric as a dependency, it seems that the byte buddy used by paparazzi cannot rewrite the Android View class. Is this the intended behavior? If not, I'm thinking of creating an issue.

class redefinition failed: attempted to change the schema (add/remove fields)

@RunWith(AndroidJUnit4::class)
class ExampleUnitTest {
    @get:Rule
    val paparazzi = Paparazzi()

It currently does not. Wanna file an issue so that it is tracked somewhere?

Thanks. I filed the issue for the time being. https://github.com/cashapp/paparazzi/issues/425