Support return statements inside switch and try/catch/finally

[marten-de-vries]

Currently working in:

• loops
• switch statements
• try/catch/finally

(a.k.a. still some work to do)

[nolanlawson]

To be clear: this is only for returns inside of switches and try/catch/finally, right? Seems to work when you don't return from inside those blocks. (Love this tool, BTW!)

[marten-de-vries]

That's right. If you encounter the issue, you can just work around it by assigning to a variable, then returning that variable after the block. That's pretty much what Kneden would do automatically for you as soon as this is fixed anyway...

[marten-de-vries]

@nolanlawson The newly opened #21 makes me less confident about my previous post. Consider Kneden an alpha-quality tool that will likely require workarounds, for now. So pretty much what the README file says.

[raphaelokon]

Is the development on this stale? Wondering how I could help out to make it non-alpha.

[marten-de-vries]

@raphaelokon Yes, development pretty much stalled. I would be open for PRs and the like, but currently I don't have the time to work on it myself. That might still change, but no guarantees.

[raphaelokon]

@marten-de-vries Thanks. I will give it a go and see if I can get the desired behavior implemented. kneden is by far the most readable and concise solution re async/await transpilation.

Are there already tests added that describe the two subtasks mentioned in this issue?
I can see all tests passing so far.

[timdp]

As an attempt to reignite this project ... To handle return inside try/catch/finally, my gut feeling would be to replace try { } blocks with _try((_return, _catch, _finally) => { }) and then replace every return inside with a call to _return that takes care of the state. I'm guessing that's already been tried though. What are the main issues? Perhaps someone can weigh in?

[ljharb]

Would my proposal and shim help here? https://npmjs.com/promise.prototype.finally

[ljharb]

@marten-de-vries are you still maintaining this? Airbnb would love to use it, but it either needs to handle all the edge cases, or throw early when attempting to transpile them, so that there's zero chance that incorrect code can make it to production.

[marten-de-vries]

@timdp Don't forget that returns also change the control flow. Still, something like that might work, and if it works I'd be fine merging something like that.
@ljharb Quite possible that would make it easier. Someone would need to try, though. As of currently, no, I'm not working on this, and chances of me picking it up again get increasingly smaller. Sorry. Implementing some easier cases and throwing on the edge ones might be easier, but again, someone would need to try it.

[trusktr]

@ljharb Can you convince AirBnB to pay for someone to fix this (at least by adding build-time failure)?