<Storytale> is now usable. You can start <here>
# feed
p
Storytale is now usable. You can start here
t
This looks like your fork of storytale, which was "usable" already. So I have 3 questions: 1. What does your fork add/improve 2. What do you mean by "is now usable" 3. Why should we not just use the upstream storytale lib
a
If you faced a bug and already fixed it in fork - please, consider to make a PR to the original repo (so we can publish new vesrion)
👍 1
p
@Tanish Azad Totally fair question! It’s easy to assume upstream works out of the box, but here’s the practical reality if you try to use it today: 1. Why not just use upstream? Upstream doesn't have a working release for modern projects. It was never published to Maven Central or the Gradle Plugin Portal, and the old dev build published to JetBrains Space (0.0.4-alpha01+dev19) is from late 2025. If you try to use it with current Kotlin (2.3+) or AGP, the compiler crashes immediately because the plugin's compiled binaries are incompatible with modern compiler APIs. Newer commits on GitHub were never published anywhere public. 2. What does this fork improve? • Available on Maven Central: You can just apply id("io.github.aryapreetam.storytale") version "0.0.7" without needing custom repos or compiling from source. • Modern toolchain support: Fully updated and tested on Kotlin 2.3/2.4, Compose Multiplatform 1.10–1.12, and AGP 9. • Works on Android Library modules: Upstream only worked on standalone Android app modules. In real design systems, UI code lives in shared library modules (com.android.library), which our fork supports out of the box. • Multiplatform runner fixes: Fixes iOS simulator crashes on Apple Silicon / Xcode 16 and makes Web/Wasm work regardless of what your module is named. • Clean build classpath: Doesn't force or leak its own AGP version onto your project. 3. What does "now usable" mean? Upstream was a great proof-of-concept, but only worked under very specific, older setups if built locally. "Now usable" means you can add it to an existing, up-to-date Compose Multiplatform project today and have your component gallery actually compile and run across Desktop, Web, Android, and iOS. We kept upstream's exact package names and API (org.jetbrains.compose.storytale), so if JetBrains resumes active releases, switching back or upstreaming these fixes will be seamless. If anyone has actually managed to get upstream resolving and running out of the box on a modern Kotlin 2.3+ setup, I’d genuinely love to see how you configured it! But for anyone hitting these same toolchain walls, 0.0.7 is there to get you unblocked.
g
@Preetam Why did you not open a PR to the original project so the jetbrains team can integrate you fixes ? (edit: you send your answer at the exact same time!)
p
@Artem Kobzar 100%, would love to contribute these back! I can prepare one consolidated PR with all the toolchain fixes to keep it simple, or split them into smaller, focused PRs if you'd prefer—whatever works best for your review workflow! Rough breakdown of what's fixed in the fork: 1. Compiler Plugin: Fixed story detection in Kotlin K2 (so stories using val x by story aren't missed) and made plugin registration compatible across Kotlin 2.3 and 2.4. 2. Android Library Support: Dynamic runner APK generation and variant resolution for
com.android.library
modules (and AGP 9.1 compatibility). 3. iOS Simulator: Detection and reuse of booted simulators on Xcode 16+ / Apple Silicon to avoid
simctl 403
errors. 4. Web / Wasm: Dynamic script name resolution instead of hardcoded
composeApp.js
. 5. Gradle Plugin Dependencies: Converted AGP/KGP to
compileOnly
so consumer projects don't leak build-tool classpaths. Let me know which format you prefer and I'll get it submitted!
@Guyaume Tremblay I originally made these fixes just to unblock my own Compose Multiplatform projects a few months back when upstream seemed quiet without public Maven Central releases. I didn't realize so many others in the community were hitting the exact same blockers until I joined #storytale channel few days back!
👍 1
t
why is publishing on maven central a requirement? i would assume rolling a private maven repository is ideal, especially for bigger orgs like jetbrains
p
Private/dev repos are great for internal CI and nightly builds, but for public tools, Maven Central is the ecosystem standard: 1. Zero Friction DX: One-line setup without plumbing custom URLs into settings.gradle.kts. 2. Enterprise Security: Most corporate environments block external Maven endpoints and only whitelist Maven Central/Google through internal proxies. 3. Reliability: Dev endpoints can sunset or move. Maven Central is permanent. All official JetBrains public libraries (Compose, Coroutines, Ktor) live on Maven Central for these exact reasons!
🧠 1
🚮 1
a
@Preetam it would be nice to have separate unit of work for each of the described problem
And yep would happy to hear more about your usage of it ❤️ Overall, thank you so much for bringing all of it together. P.S. me and @Oleksandr Karpovich [JB] are discussing the possibility to publish the first version to the public Maven because of your points!
kodee happy 6
👍 1
p
Thank you @Artem Kobzar! Having Storytale on Maven Central will make adoption much easier across the community. Separate PRs make complete sense. I’ll prepare clean, isolated branches against
main
for each issue so they're easy to review and merge one by one. I'll get the first one (K2 compiler plugin fixes) submitted shortly. For context on my usage: I was working on a Compose Multiplatform project with a lot of complex custom components. Maintaining them across different targets was tough, and Storytale made it much easier to develop and tweak them in isolation. I have also built an e2e testing library for CMP (Parikshan) and wanted to enable something like
@storybook/test
to prevent regressions in component UI and behavior across Desktop, Android, iOS, and Web. That brought me back to Storytale. When I saw folks on Slack running into the same toolchain blockers I had worked through, I decided to bundle the fixes and publish them so they work out of the box. Excited to get these merged into upstream!
kodee excited 2
Hey @Artem Kobzar, as suggested, I've started splitting the fixes into isolated, minimal units of work. Here is the first PR focusing strictly on the compiler plugin: github.com/Kotlin/Storytale/pull/116 Summary: • Matches story property delegates directly by factory type (fixes K2 discovering 0 stories). • Safe-casts FIR metadata source to avoid ClassCastException. • Bumps kotlinCompileTesting to 0.12.1 to resolve the test harness failure on Kotlin 2.3+. Verified against the examples module on both Desktop and Wasm. Whenever you and @Oleksandr Karpovich [JB] get a chance to take a look, would love your feedback!
Hi @Artem Kobzar, @Oleksandr Karpovich [JB]— thanks again for reviewing and merging #116! 🙌 Following Artem’s suggestion to keep follow-ups small and atomic, I’ve opened the next PR focusing just on the Gradle plugin’s Web/Wasm script resolution: github.com/Kotlin/Storytale/pull/117 Right now
composeApp.js
is hardcoded in the generator tasks, which leads to 404s when Storytale is applied to modular project layouts (like
:shared
or
:app:shared
). This PR dynamically resolves the output bundle name and includes
ProjectBuilder
unit tests covering the common directory layouts. I figured it made sense to submit this self-contained fix first before diving into the heavier Android library / iOS runner setups.
👀 1
thank you color 1
Thanks a lot for merging #117, @Artem Kobzar! Whenever you get some time, could you please take a look at PR #118 (Android library support)? I’ve rebased it on the latest main, and CI is green.
thank you color 1
Hi @Artem Kobzar @Oleksandr Karpovich [JB], To give you visibility into what I'm working on and how the pieces fit together, here is our planned PR pipeline: 1. PR #118 (Current): Android Library Support — enables Storytale for projects using
com.android.library
alongside AGP 8.x/9.x. 2. Next PR: Compiler Plugin K2 ABI Compatibility — ensures Storytale compiler plugin works seamlessly when consumed by modern projects on Kotlin 2.4+ / CMP 1.11+ without runtime
ClassCastException
in
ExtensionStorage
. 3. Follow-up PR: iOS Storytale Runner enhancements — stabilizing native task execution and resource packaging for iOS targets. 4. Follow-up PR: Documentation, example use-cases, and gallery-demo refactoring. I am running local verification on top of latest
main
now and will update PR #118 shortly once all checks pass. Let me know if that works!
🙌 1
thank you color 1
PR #118 is now rebased on latest main with all local checks and physical device runs passing across application and KMP library targets. Ready for review!
Hey @Artem Kobzar, @Oleksandr Karpovich [JB], just following up on PR #118 when you have a moment. Happy to adjust the implementation or break it into smaller PRs if you prefer.