A modern programming language that makes developers happier.

kotlinlang

image.png

*Edit:* Solution in thread :thread:

Was encouraged to bring this here from another slack

I am once again wondering if I understand how coroutines work and it's an issue that's stumped a few of us :sweat_smile:

Trying to add basic parallelization using a parallel mapping function in <https://github.com/slackhq/slack-gradle-plugin/pull/678>

1. Create new thread pool context: <https://github.com/slackhq/slack-gradle-plugin/pull/678/files#diff-a8b4c979f21b3d6249822cd302bc12879d634c30c6c0ddbbd527d6d3299841e1R125>
2. that context is passed to the runner: <https://github.com/slackhq/slack-gradle-plugin/pull/678/files#diff-a8b4c979f21b3d6249822cd302bc12879d634c30c6c0ddbbd527d6d3299841e1R117>
3. context is used in `withContext()` <https://github.com/slackhq/slack-gradle-plugin/pull/678/files#diff-8b5029da572289698d8ca52c5bfb70e1c53262375b55c494bbfe916db5e9adc1R75>
4. Each rail runs on the same thread despite the context's dispatcher being backed with a n=size thread pool :thinking_face:. Same behavior happens with `<http://Dispatchers.IO|Dispatchers.IO>` and `Dispatchers.Default` too
The parallel mapper does work internally as expected (with some fun tests), so I don't think it's an issue with that function directly

Solved: problem was the use of `UNDISPATCHED` in the parallel mapper, which ran on the same thread until a suspension point. As the function called in the body wasn't a suspend function, no suspension point ever occurred