segmentation fault in recursive Task

[zubchick]

Hi, I tried to reproduce python code in julia

In [1]: def f():
   ...:     yield f()

In [2]: next(f())
Out[2]: <generator object f at 0x1427640>

In [3]: next(next(f()))
Out[3]: <generator object f at 0x1427730>

I'm not really sure that i wrote it right:

julia> function test()
       produce(test())
       end

julia> t = Task(test)
Task

julia> consume(t)

And got this:

julia(90886,0x7fff73df5960) malloc: *** mmap(size=140730481774592) failed (error code=12)
*** error: can't allocate region
*** set a breakpoint in malloc_error_break to debug
julia(90886,0x7fff73df5960) malloc: *** mmap(size=140730481782784) failed (error code=12)
*** error: can't allocate region
*** set a breakpoint in malloc_error_break to debug
julia(90886,0x7fff73df5960) malloc: *** mmap(size=140730481790976) failed (error code=12)
*** error: can't allocate region
*** set a breakpoint in malloc_error_break to debug
^C^C^C
[1]    90886 segmentation fault  ./julia

Version 0.0.0+100556114.r27bb

[toivoh]

In python, any function that uses yield automatically returns a generator.
In julia, you always have to wrap the function into a Task manually, so your code should be

function test()
    produce(Task(test))
end

which works for me.

The function

function test()
    produce(test())
end

evaluates itself in an infinite recursion trying to produce a value to give to produce, so it should give a stack overflow error, and it does if I call it as test(). I can confirm that your code gives a segfault for me too, so yes, this must be a bug.

[JeffBezanson]

The stack overflow is certainly the origin of the problem.