Python: memory leak with memory_profiler

Question

I want to use memory_profiler package to analyze memory usage.

However, I've some confusions:

Example 1

# test1.py
from memory_profiler import profile

class Object:
    def __init__(self):
        pass


list = []

@profile
def func():
    for i in range(100000):
        list.append(Object())


func()

Results 1

First test

mprof: Sampling memory every 0.1s
running new process
running as a Python program...
Filename: test1.py

Line #    Mem usage    Increment  Occurrences   Line Contents
=============================================================
    10     43.9 MiB     43.9 MiB           1   @profile
    11                                         def func():
    12     50.1 MiB      0.0 MiB      100001       for i in range(100000):
    13     50.1 MiB      6.2 MiB      100000           l.append(Object())

Second test

mprof: Sampling memory every 0.1s
running new process
running as a Python program...
Filename: test1.py

Line #    Mem usage    Increment  Occurrences   Line Contents
=============================================================
    10     44.2 MiB     44.2 MiB           1   @profile
    11                                         def func():
    12     50.3 MiB  -2867.4 MiB      100001       for i in range(100000):
    13     50.3 MiB  -2861.3 MiB      100000           l.append(Object())

Question 1

About test1.py:

1. The result of memory usage is different some time.
2. Are there have memory leak in this example? I think python have garbage collection mechanism, so list should be None which will not cause memory leak. However, the result of mprof plot show usage increase.

---

Example 2

# test2.py
from memory_profiler import profile

class Object:
    def __init__(self):
        pass


@profile
def func():
    global list
    list = []
    for i in range(100000):
        list.append(Object())


func()

Results 2

mprof: Sampling memory every 0.1s
running new process
running as a Python program...
Filename: test2.py

Line #    Mem usage    Increment  Occurrences   Line Contents
=============================================================
     7     44.2 MiB     44.2 MiB           1   @profile
     8                                         def func():
     9                                             global l
    10     44.2 MiB      0.0 MiB           1       l = []
    11     50.3 MiB  -2905.3 MiB      100001       for i in range(100000):
    12     50.3 MiB  -2899.2 MiB      100000           l.append(Object())

Question 2

About test2.py:

1. I think this example will cause a memory leak. The reason is that after the function ends, list as a global variable still holds a reference to the list, and the objects in the list are not released, resulting in a memory leak.
2. But why the result of Increment was -2899.2 MiB which seem there are not happend memory leak.