Hello! I created two issues with some proposals to Koin. Can you provide some feedback, please. <Iss...

# koin-contributors

[Alexander Zhdanov]

Hello! I created two issues with some proposals to Koin. Can you provide some feedback, please. Issue one and issue two.

[arnaud.giuliani]

yes, can be good to make bind DSL stronger 👍

[Alexander Zhdanov]

Making bind overload with KClass parameter type safe is not backward-compatible, so it can be done in 3.x release

[Alexander Zhdanov]

But bind overload without KClass parameter can't be made type-safe without extra verbocity in its usage 😟

[arnaud.giuliani]

let’s check that

[Alexander Zhdanov]

After some investigation I found, that checking KClass of BeanDefinition can be done to ensure type safety, but kotlin's reflection is not available in commonMain module.

[arnaud.giuliani]

yeah, I would avoid kotlin-reflection in main

[Alexander Zhdanov]

I created MR to make all other bind methods safe. IMHO, bind without params may be deprecated, until checking supertype of KClass will be available without reflection

[arnaud.giuliani]

yep, thanks

[Alexander Zhdanov]

I have added type check for BeanDefinition.secondaryTypes in checkModules, so types of bind without params now may be checked at compile time by CheckModulesTest.

[arnaud.giuliani]

checkModules is not covering enough. I believe the key will be around plugin compiler analysis

[Alexander Zhdanov]

yes, but such checkModules is useful now, when plugin compiler is not available yet

[arnaud.giuliani]

for now yes

[Fabio]

@Alexander Zhdanov hey, I just had a look at your @inject ticket. I've been keen on having that too.
I'm thirsty to use ksp (as in #ksp channel) to enhance checkModules first though, through code generation.
I see them as related because one of the ways to do @inject would be generate the graph and code at compile time. Not a small change though.

[arnaud.giuliani]

let’s open a thread on #koin-dev 👍 would be great to start this topic