Make recursing too deep a fatal runtime error and make the limit larger

[brson]

What we are doing now leaks (#3555) and I don't know how to fix it yet, so we should not pretend we can enforce stack size limits safely.

[eholk]

Hmm... limits on recursion depth make me cringe.

Why have stack limits at all if we can just keep growing the stack?

[brson]

As a sanity check. Eating all your ram and having the machine lock up while it swaps and oom kills is no fun. Nobody means to recurse infinitely.

[brson]

FWIW I tested infinite recursion on Go over the weekend and let the process grow to over 1GB before I killed it manually. Not sure if they just recurse forever or not.

[DonMartio]

Disregarding the risk of being off topic. Are there any plans implementing tail recursion?

[catamorphism]

@DonMartio - We're not sure yet; see #217.

[brson]

gcc's __morestack calls abort if allocation fails

[nikomatsakis]

Not critical for 0.6; removing milestone

[catamorphism]

Nominating for milestone 5, production-ready

[graydon]

accepted for production-ready milestone

[jbclements]

+1 to eric holk's comment, basically.

[emberian]

Visiting for triage; nothing to add.

[eholk]

Is this bug still open after 149047e?

If so, I'm still not convinced it's a good idea. After all, you can make the machine grind to a halt swapping by allocating lots of memory from the heap, and yet we don't put arbitrary limits on the heap size.

[alexcrichton]

Closing in favor of #11011 which outlines the strategy we should be taking and is a little more up-to-date.