I like where you're going with this. I'm guessing in this case you specifically don't want the error to prevent the metric from being parsed? or are you just trying to make the error visible?

@ssoroka

I like where you're going with this. I'm guessing in this case you specifically don't want the error to prevent the metric from being parsed? or are you just trying to make the error visible?

I would answer both, I want to have the full control if an error occurs. But initially my use case was indeed to ensure that the metric is processed in case of an error. With the current code, if you have an error, the metric is lost while I would like to keep it and add the error message to it.

I agree the user should have full control over what happens. I was thinking there should be some logging primitives, too, to tie into the Telegraf logger. something like log.info, log.warn, log.error, and log.debug, so you can customize what to log around error conditions.

I'd also like the ability to explicitly return an error. I believe starlark has a built-in fail() command that can do this, but I haven't tested it.

@ssoroka

I'd also like the ability to explicitly return an error. I believe starlark has a built-in fail() command that can do this, but I haven't tested it.

Yes the fail function is part starlarktest and is for testing purpose. It actually leverages the (same) catch function as you can see here

Have any interest in adding the other functions I mentioned in another PR? I've got a list of outstanding starlark issues here, and I'd be more than happy for the assistance. area/starlark

also feel free to join the #Telegraf channel on InfluxData slack, and consider writing starlark-based tests in the future as they get executed and double as documentation

Have any interest in adding the other functions I mentioned in another PR? I've got a list of outstanding starlark issues here, and I'd be more than happy for the assistance. area/starlark

Thanks for proposing, yes I'm interested in helping you if I can.

I will try to propose a PR asap for the logging, I guess it is #8402

feel free to add to that issue. it's mostly just a placeholder to attach work and descriptions to later.