Limit max recursion depth by default.

[kunom]

Related to #1192.

[jdhardy]

Unfortunately the RecursionLimit is basically ignored. You would also need to track every python stack frame entry/exit and break when the limit is hit. It's probably possible but I'm not sure what the performance hit might be.

[DinoV]

The reason why this is disabled by default is due to performance issues and there'll be a significant hit to function call performance from enabling this by default. I don't think the limit is ignored though currently, we should do the proper enforcement every time a function is called, and someone hitting the issue here can always run with -X:RecursionLimit set

[kunom]

jdhardy, I am sorry but I need some more time to be able to follow your comment. At least in my tests with hasattr, setting a recursion limit worked.

DinoV, can you elaborate a bit more on what the exact significant performance issues are with enforcing the recursion limit? Is it the counter management in PythonFunction.AddRecursionDepth()? From a naive standpoint, enforcing the recursion limit is not more than incrementing, decrementing and comparing an integer.

Regarding #1192, I don't see an other way to achieve CPython compatiblity since StackOverflowExceptions are normally uncatchable since .NET 2.0. (In fatch, there is already a catch-all handler in PythonOps.HasAttr()).

[slide]

If you would like to recreate this PR, please do so under IronLanguages/ironpython2