Is there are reason to prefer one or the other of `Json.decodeFromString(file.readText())` vs. `file...

# serialization

[Sebastian Schuberth]

Is there are reason to prefer one or the other of Json.decodeFromString(file.readText()) vs. file.outputStream().use { Json.decodeFromStream(it) }?

[ephemient]

given what kotlinx.serialization has been optimized for, I would expect the first to perform better. you can do your own benchmarking

[ephemient]

https://github.com/Kotlin/kotlinx.serialization/issues/2186

[Sebastian Schuberth]

Ah, interesting, thanks for that pointer!

[Sebastian Schuberth]

However, from skimming over the issue it seems to have been addressed with version 1.5.0, so there is no performance difference anymore?

[ephemient]

run your own benchmarks. I'm pretty sure it's still slower than Moshi whose bytes-to-UTF-8 path has been much more used and optimized

[Sebastian Schuberth]

I'm not really interested in a comparison to Moshi, but in a KxS string vs stream API comparison. Do you have any pointers to benchmarks for that?

[CLOVIS]

Do you have any pointers to benchmarks for that?

https://github.com/Kotlin/kotlinx-benchmark

If you do write your own microbenchmark though, I highly recommend posting it here, because it's notoriously easy to write them wrong, so you'll want it to be reviewed by other people

[Sebastian Schuberth]

Thanks, but I'll most probably not go into the rabbit hole of writing my own benchmark 😉 I was more looking for a "serious" source to refer to.