Updating the description of Future exception re-throwing

[demobox]

See https://groups.google.com/forum/#!topic/scala-language/PeAuAK6_Aj8

/cc @viktorklang

[SethTisue]

I believe you want to be updating http://docs.scala-lang.org/overviews/core/futures.html, not the SIP page?

[demobox]

Hi @SethTisue

I believe you want to be updating http://docs.scala-lang.org/overviews/core/futures.html

Yes, er, that was indeed the plan ;-) Of the five pages I could find that obviously reference futures:

I thought this one (overviews/core/_posts/2012-09-20-futures.md) was the most likely. If it's not the right one, please let me know which doc I should be updating instead.

Thanks!

[SethTisue]

You're totally right. I was confused.

[viktorklang]

Nice. I think it would be great to also explain the rationality for rethrowing Fatals (fail fast).

[demobox]

the rationality for rethrowing Fatals (fail fast).

@viktorklang Could you explain that in a bit more detail? Unless I'm misunderstanding something, it's not the caller that fails fast, it's the worker thread that was actually executing the body of the Future.

Do you mean that the idea is to ensure that worker processes that are "irretrievably broken" die off as soon as a non-recoverable error is detected? In that case, how about the following?

"Fatal exceptions (as determined by NonFatal) are rethrown in the thread executing the failed asynchronous computation.This ensures that "broken" threads in the ExecutionContext fail fast, and can be replaced if necessary. See NonFatal for a more precise description of the semantics."

[viktorklang]

It's rethrown to the worker to escalate the problem upwards (instead of pushing the problem downstream to the consumer, which may or may not be listening or dealing with those kinds of problems).

[demobox]

@viktorklang Thanks for clarifying. In that case, how about:

"Fatal exceptions (as determined by NonFatal) are rethrown in the thread executing the failed asynchronous computation.This informs the code managing the executing threads of the problem and allows it to fail fast, if necessary. See NonFatal for a more precise description of the semantics."

?

[viktorklang]

SGTM

[demobox]

@viktorklang Updated. Let me know if you'd like me to squash the commits down.

Thanks!

[viktorklang]

That's up to @SethTisue :)

[demobox]

That's up to Seth Tisue :)

@SethTisue: ping..? ;-)

[SethTisue]

to my knowledge, we're not picky about that in this repo.

[SethTisue]

thank you, @demobox! much appreciated.

[demobox]

SethTisue merged commit 3870bc3

Thanks, @SethTisue!