Read kotlin metadata to resolve nullability of suspending functions

[ferinagy]

Attempt at #3075. Any suggestions to improve are welcome.

[ZacSweers]

Why not use kotlinx-metadata?

[cbruegg]

@ZacSweers Due to its size: #3075 (comment)

[ZacSweers]

That's two years old and almost certainly referring to kotlin-reflect, which is different.

[cbruegg]

It’s still 2.6 MB, right? https://mvnrepository.com/artifact/org.jetbrains.kotlinx/kotlinx-metadata-jvm/0.2.0

[ZacSweers]

Sure, but that's not what the comment you linked was referring to.

[ferinagy]

Why not use kotlinx-metadata?

It's in the name of the issue #3075: "...and avoid kotlin-metadata-jvm dep". 🤷

That is actually how I started, I can fairly easily put it back. For me it shows closer to 2.7MB. It probably would be stripped by proguard/r8 a lot, since we really need just a small bit, but still.

[JakeWharton]

The metadata dependency is unacceptable, yes. I'll look later. Thanks for the PR.

[ferinagy]

Hi, no pressure here, but if I can explain or clean up the code to help with reviewing this, let me know.

[JakeWharton]

Earliest I'll be able to look is next week. I'm just very busy, sorry!

[Mehly]

Any update on this? :)

[ferinagy]

I did another pass through this. I cleaned up some stuff, marked which parts were extracted from kotlinx-metadata-jvm and kotlinx.serialization, and added some comments to the rest. Any feedback will be appreciated.

[krokyze]

Any update on this? :)

Looks like kotlinx-metadata has a newer version which is now 1MB in size https://mvnrepository.com/artifact/org.jetbrains.kotlinx/kotlinx-metadata-jvm/0.4.0

[pm-nchain]

Any updates on this? :)